diff --git a/article/VRostami_JIto_MDenker_SGruen_2016.bib b/article/Rostami-Ito-Denker-Gruen-2017.bib similarity index 100% rename from article/VRostami_JIto_MDenker_SGruen_2016.bib rename to article/Rostami-Ito-Denker-Gruen-2017.bib diff --git a/article/VRostami_JIto_MDenker_SGruen_2016.md b/article/Rostami-Ito-Denker-Gruen-2017.md similarity index 98% rename from article/VRostami_JIto_MDenker_SGruen_2016.md rename to article/Rostami-Ito-Denker-Gruen-2017.md index f509b9a..57adc19 100644 --- a/article/VRostami_JIto_MDenker_SGruen_2016.md +++ b/article/Rostami-Ito-Denker-Gruen-2017.md @@ -19,21 +19,22 @@ Address: Contact: - v.rostami@fz-juelich.de Editor: - - Name Surname + - Nicolas P. Rougier Reviewer: - - Name Surname - - Name Surname + - Georgios Detorakis + - Fabien Benureau Publication: - received: Nov, 1, 2016 - accepted: Nov, 1, 2016 - published: Nov, 1, 2016 - volume: "**1**" + received: Oct 28, 2016 + accepted: Apr 6, 2017 + published: May 29, 2017 + volume: "**3**" issue: "**1**" - date: Nov 2016 + number: 3 + date: May 2017 Repository: - article: "http://github.com/rescience/rescience-submission/article" - code: "http://github.com/rescience/rescience-submission/code" - data: + article: "https://github.com/ReScience-Archives/Rostami-Ito-Denker-Gruen-2017/tree/master/article" + code: "https://github.com/ReScience-Archives/Rostami-Ito-Denker-Gruen-2017/tree/master/code" + data: "https://github.com/ReScience-Archives/Rostami-Ito-Denker-Gruen-2017/tree/master/data" notebook: Reproduction: - "Spike synchronization and rate modulation differentially involved in motor cortical function. Alexa Riehle, Sonja Grün, Markus Diesmann, and Ad Aertsen (1997) Science 278:1950-1953. DOI:10.1126/science.278.5345.19 diff --git a/article/VRostami_JIto_MDenker_SGruen_2016.bbl b/article/VRostami_JIto_MDenker_SGruen_2016.bbl deleted file mode 100644 index b98a80b..0000000 --- a/article/VRostami_JIto_MDenker_SGruen_2016.bbl +++ /dev/null @@ -1,594 +0,0 @@ -% $ biblatex auxiliary file $ -% $ biblatex bbl format version 2.5 $ -% Do not modify the above lines! -% -% This is an auxiliary file used by the 'biblatex' package. -% This file may safely be deleted. It will be recreated by -% biber as required. -% -\begingroup -\makeatletter -\@ifundefined{ver@biblatex.sty} - {\@latex@error - {Missing 'biblatex' package} - {The bibliography requires the 'biblatex' package.} - \aftergroup\endinput} - {} -\endgroup - - -\refsection{0} - \sortlist{nty}{nty} - \entry{Abeles82}{book}{} - \name{author}{1}{}{% - {{hash=47cc27c6ef754465fea8a3274304d0f3}{Abeles}{A\bibinitperiod}{Moshe}{M\bibinitperiod}{}{}{}{}}% - } - \list{location}{1}{% - {Berlin, Heidelberg, New York}% - } - \list{publisher}{1}{% - {Springer-Verlag}% - } - \strng{namehash}{47cc27c6ef754465fea8a3274304d0f3} - \strng{fullhash}{47cc27c6ef754465fea8a3274304d0f3} - \field{sortinit}{A} - \field{sortinithash}{b685c7856330eaee22789815b49de9bb} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{series}{Studies of Brain Function} - \field{title}{{Local Cortical Circuits: An Electrophysiological Study}} - \field{year}{1982} - \endentry - \entry{Adrian2014}{article}{} - \name{author}{5}{}{% - {{hash=1b6c773c24479f4228162b5adabae6f7}{Adrian}{A\bibinitperiod}{Stoewer}{S\bibinitperiod}{}{}{}{}}% - {{hash=a38bc839f9ea464ff7351b7971bb1d40}{Christian}{C\bibinitperiod}{Kellner}{K\bibinitperiod}{}{}{}{}}% - {{hash=51682a24e542a40a2789ebaf30f8bbc8}{Jan}{J\bibinitperiod}{Benda}{B\bibinitperiod}{}{}{}{}}% - {{hash=0bec959f3fd1708a0a22ac29a8d89aed}{Thomas}{T\bibinitperiod}{Wachtler}{W\bibinitperiod}{}{}{}{}}% - {{hash=c9526a3dcdcbb4cbfcf729535d58d47e}{Jan}{J\bibinitperiod}{Grewe}{G\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{ed89cc06de4e84b91e7955eceb3a7907} - \strng{fullhash}{2824824ef03e8305c553a3408e5b2d27} - \field{sortinit}{A} - \field{sortinithash}{b685c7856330eaee22789815b49de9bb} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Frontiers in Neuroinformatics} - \field{title}{{File format and library for neuroscience data and metadata}} - \field{volume}{8} - \field{year}{2014} - \endentry - \entry{badia_incf_2015}{report}{} - \name{author}{16}{}{% - {{hash=243314f3a9e4f580132d3fd8619f8c55}{Badia}{B\bibinitperiod}{Rosa}{R\bibinitperiod}{}{}{}{}}% - {{hash=72b1f79a6f35439abb27d2ca5236fb0b}{Davison}{D\bibinitperiod}{Andrew}{A\bibinitperiod}{}{}{}{}}% - {{hash=f6f72f78b262d4fcff83ad67927cd29c}{Denker}{D\bibinitperiod}{Michael}{M\bibinitperiod}{}{}{}{}}% - {{hash=59f22b302d98e9e075ec814f17c1e49a}{Giesler}{G\bibinitperiod}{Andre}{A\bibinitperiod}{}{}{}{}}% - {{hash=6a3efe5c8970888f8396d808b125b35d}{Gosh}{G\bibinitperiod}{Satra}{S\bibinitperiod}{}{}{}{}}% - {{hash=fae216475918e7d5a3fd11e334229e70}{Goble}{G\bibinitperiod}{Carole}{C\bibinitperiod}{}{}{}{}}% - {{hash=23e27f78671208ddad6a55ffec7d1cce}{Grewe}{G\bibinitperiod}{Jan}{J\bibinitperiod}{}{}{}{}}% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - {{hash=147826d25121323a650d728dcf00c29f}{Hatsopoulos}{H\bibinitperiod}{Nicho}{N\bibinitperiod}{}{}{}{}}% - {{hash=94c91bf485ff901c4b7791cbf12eb9a4}{LeFranc}{L\bibinitperiod}{Yann}{Y\bibinitperiod}{}{}{}{}}% - {{hash=8eda2b6e97bca9218ea6b70dcb8cd32c}{Muller}{M\bibinitperiod}{Jeff}{J\bibinitperiod}{}{}{}{}}% - {{hash=ba464751e85c5975fd4b00830dce739e}{Pröpper}{P\bibinitperiod}{Robert}{R\bibinitperiod}{}{}{}{}}% - {{hash=c757ce20c6904e9c0e5b93c611bc0992}{Teeters}{T\bibinitperiod}{Jeff}{J\bibinitperiod}{}{}{}{}}% - {{hash=739fea8110d6909bf6645f56b5fc629c}{Wachtler}{W\bibinitperiod}{Thomas}{T\bibinitperiod}{}{}{}{}}% - {{hash=63fafe1a09fb369e618094cf3225c974}{Weeks}{W\bibinitperiod}{Michael}{M\bibinitperiod}{}{}{}{}}% - {{hash=86a847fed233df0376116790ede82340}{Zehl}{Z\bibinitperiod}{Lyuba}{L\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{686d32af40b9ae39bbbaa2b1d5e4bc84} - \strng{fullhash}{1583fd2fa12ebe3892fb1c6003c4e90c} - \field{sortinit}{B} - \field{sortinithash}{4ecbea03efd0532989d3836d1a048c32} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{title}{{INCF} {Program} on {Standards} for data sharing: new perspectives on workflows and data management for the analysis of electrophysiological data.} - \field{type}{Techn. {Report}} - \field{year}{2015} - \endentry - \entry{Brown04_456}{article}{} - \name{author}{3}{}{% - {{hash=e4cbf6c251a42f3473ec76634f8e570d}{Brown}{B\bibinitperiod}{Emery\bibnamedelima N.}{E\bibinitperiod\bibinitdelim N\bibinitperiod}{}{}{}{}}% - {{hash=5608b50bd0c7936040f0c7d3836d31e3}{Kaas}{K\bibinitperiod}{Robert\bibnamedelima E.}{R\bibinitperiod\bibinitdelim E\bibinitperiod}{}{}{}{}}% - {{hash=bd7f0337a0a5a73e8dab8370e4deac89}{Mitra}{M\bibinitperiod}{Partha\bibnamedelima P.}{P\bibinitperiod\bibinitdelim P\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{642fb0f81a9deb0604f09c754890f302} - \strng{fullhash}{642fb0f81a9deb0604f09c754890f302} - \field{sortinit}{B} - \field{sortinithash}{4ecbea03efd0532989d3836d1a048c32} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Nature Neuroscience} - \field{title}{Multiple neural spike train data analysis: state-of-the-art and future challenges} - \field{volume}{7} - \field{year}{2004} - \field{pages}{456\bibrangedash 461} - \range{pages}{6} - \endentry - \entry{Denker2015_000}{inbook}{} - \name{author}{2}{}{% - {{hash=f6f72f78b262d4fcff83ad67927cd29c}{Denker}{D\bibinitperiod}{Michael}{M\bibinitperiod}{}{}{}{}}% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - } - \name{editor}{4}{}{% - {{hash=b393ec8ba7421eb4f3c1f05e1bb6fd9d}{Amunts}{A\bibinitperiod}{Katrin}{K\bibinitperiod}{}{}{}{}}% - {{hash=8a5319fcc8c034f02d1e6131a2075556}{Grandinetti}{G\bibinitperiod}{Lucio}{L\bibinitperiod}{}{}{}{}}% - {{hash=fdb7d4a16eb3f3118ac7a8e12f2719ce}{Lippert}{L\bibinitperiod}{Thomas}{T\bibinitperiod}{}{}{}{}}% - {{hash=eef9a82a22125c381c2932d53b56b78c}{Petkov}{P\bibinitperiod}{Nicolai}{N\bibinitperiod}{}{}{}{}}% - } - \list{publisher}{1}{% - {Springer Series Lecture Notes in Computer Science}% - } - \strng{namehash}{a1aa963d64c6fcf456bbb3ce1075c91c} - \strng{fullhash}{a1aa963d64c6fcf456bbb3ce1075c91c} - \field{sortinit}{D} - \field{sortinithash}{78f7c4753a2004675f316a80bdb31742} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{chapter}{Designing Workflows for the Reproducible Analysis of Electrophysiological Data.} - \field{title}{Brain Inspired Computing} - \field{year}{in press} - \endentry - \entry{Denker10}{article}{} - \name{author}{4}{}{% - {{hash=f6f72f78b262d4fcff83ad67927cd29c}{Denker}{D\bibinitperiod}{Michael}{M\bibinitperiod}{}{}{}{}}% - {{hash=0840237d57f02a400049f66f51b385d9}{Riehle}{R\bibinitperiod}{Alexa}{A\bibinitperiod}{}{}{}{}}% - {{hash=a6e9c8b3b1dc422f6d4f5f42ab3051bc}{Diesmann}{D\bibinitperiod}{Markus}{M\bibinitperiod}{}{}{}{}}% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{dca23b7cad1ffa842bc32d6b8d6b12b6} - \strng{fullhash}{57447a90c8ea53d4705b05a2faa34c16} - \field{sortinit}{D} - \field{sortinithash}{78f7c4753a2004675f316a80bdb31742} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Journal of Computational Neuroscience} - \field{title}{Estimating the contribution of assembly activity to cortical dynamics from spike and population measures} - \field{volume}{29} - \field{year}{2010} - \field{pages}{599\bibrangedash 613} - \range{pages}{15} - \endentry - \entry{Denker11}{article}{} - \name{author}{6}{}{% - {{hash=f6f72f78b262d4fcff83ad67927cd29c}{Denker}{D\bibinitperiod}{Michael}{M\bibinitperiod}{}{}{}{}}% - {{hash=282f98a46dceada7afac6b3ecf81dfec}{Roux}{R\bibinitperiod}{Sébastien}{S\bibinitperiod}{}{}{}{}}% - {{hash=636dfb9df02f33d2e9354d37a46a1d4d}{Lindén}{L\bibinitperiod}{Henrik}{H\bibinitperiod}{}{}{}{}}% - {{hash=a6e9c8b3b1dc422f6d4f5f42ab3051bc}{Diesmann}{D\bibinitperiod}{Markus}{M\bibinitperiod}{}{}{}{}}% - {{hash=0840237d57f02a400049f66f51b385d9}{Riehle}{R\bibinitperiod}{Alexa}{A\bibinitperiod}{}{}{}{}}% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{dca23b7cad1ffa842bc32d6b8d6b12b6} - \strng{fullhash}{1743353b31e847fab795d69a76533215} - \field{sortinit}{D} - \field{sortinithash}{78f7c4753a2004675f316a80bdb31742} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Cerebral Cortex} - \field{title}{The Local Field Potential Reflects Surplus Spike Synchrony} - \field{volume}{21} - \field{year}{2011} - \field{pages}{2681\bibrangedash 2695} - \range{pages}{15} - \endentry - \entry{Diesmann16_personalcomm}{article}{} - \name{author}{1}{}{% - {{hash=a6e9c8b3b1dc422f6d4f5f42ab3051bc}{Diesmann}{D\bibinitperiod}{Markus}{M\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{a6e9c8b3b1dc422f6d4f5f42ab3051bc} - \strng{fullhash}{a6e9c8b3b1dc422f6d4f5f42ab3051bc} - \field{sortinit}{D} - \field{sortinithash}{78f7c4753a2004675f316a80bdb31742} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{personal communication} - \field{title}{Report on comparing the IDL and the Matlab implementations of the UE analysis} - \field{year}{2016} - \endentry - \entry{Garcia14}{article}{} - \name{author}{11}{}{% - {{hash=71d56699b4609102c490a47fd1b4afb3}{Garcia}{G\bibinitperiod}{Samuel}{S\bibinitperiod}{}{}{}{}}% - {{hash=1cb60289022dc8a4246825aa4e982034}{Guarino}{G\bibinitperiod}{Domenico}{D\bibinitperiod}{}{}{}{}}% - {{hash=b124ef01d8cff212ad2f689dc37ff92d}{Jaillet}{J\bibinitperiod}{Florent}{F\bibinitperiod}{}{}{}{}}% - {{hash=328f84fba2698c187d276e52b83df523}{Jennings}{J\bibinitperiod}{Todd}{T\bibinitperiod}{}{}{}{}}% - {{hash=ba464751e85c5975fd4b00830dce739e}{Pröpper}{P\bibinitperiod}{Robert}{R\bibinitperiod}{}{}{}{}}% - {{hash=f51cd2205aa0844293cfb096b550b0b7}{Rautenberg}{R\bibinitperiod}{Philipp}{P\bibinitperiod}{}{}{}{}}% - {{hash=964cba39e729aaf947dfb9e0547e2313}{Rodgers}{R\bibinitperiod}{Chris}{C\bibinitperiod}{}{}{}{}}% - {{hash=5c54dce88d4e8d05a24fda7d6f7c898e}{Sobolev}{S\bibinitperiod}{Andrey}{A\bibinitperiod}{}{}{}{}}% - {{hash=739fea8110d6909bf6645f56b5fc629c}{Wachtler}{W\bibinitperiod}{Thomas}{T\bibinitperiod}{}{}{}{}}% - {{hash=ceebf33d0edf9c8fbb25b47b41cd2b32}{Yger}{Y\bibinitperiod}{Pierre}{P\bibinitperiod}{}{}{}{}}% - {{hash=72b1f79a6f35439abb27d2ca5236fb0b}{Davison}{D\bibinitperiod}{Andrew}{A\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{53fbb3546f55a61d2ea097998d43b3f0} - \strng{fullhash}{53772086c1df2cd16c385960112ff039} - \field{sortinit}{G} - \field{sortinithash}{1c854ef9177a91bf894e66485bdbd3ed} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Frontiers in Neuroinformatics} - \field{title}{Neo: an object model for handling electrophysiology data in multiple formats} - \field{volume}{8} - \field{year}{2014} - \field{pages}{10} - \range{pages}{1} - \endentry - \entry{Grammont99}{article}{} - \name{author}{2}{}{% - {{hash=6b07390586ee048fffae07a3e2dcee7c}{Grammont}{G\bibinitperiod}{Franck}{F\bibinitperiod}{}{}{}{}}% - {{hash=0840237d57f02a400049f66f51b385d9}{Riehle}{R\bibinitperiod}{Alexa}{A\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{f85968b288bf0ed54621169def382bad} - \strng{fullhash}{f85968b288bf0ed54621169def382bad} - \field{sortinit}{G} - \field{sortinithash}{1c854ef9177a91bf894e66485bdbd3ed} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Experimental Brain Research} - \field{title}{Precise spike synchronization in monkey motor cortex involved in preparation for movement} - \field{volume}{128} - \field{year}{1999} - \field{pages}{118\bibrangedash 122} - \range{pages}{5} - \endentry - \entry{Grewe2011}{article}{} - \name{author}{3}{}{% - {{hash=23e27f78671208ddad6a55ffec7d1cce}{Grewe}{G\bibinitperiod}{Jan}{J\bibinitperiod}{}{}{}{}}% - {{hash=739fea8110d6909bf6645f56b5fc629c}{Wachtler}{W\bibinitperiod}{Thomas}{T\bibinitperiod}{}{}{}{}}% - {{hash=f0049b5905ec2af7caf4cdf6b716a30a}{Benda}{B\bibinitperiod}{Jan}{J\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{7619fdad426faca7c44ef79277863b4c} - \strng{fullhash}{7619fdad426faca7c44ef79277863b4c} - \field{sortinit}{G} - \field{sortinithash}{1c854ef9177a91bf894e66485bdbd3ed} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Frontiers in Neuroinformatics} - \field{title}{{A Bottom-up Approach to Data Annotation in Neurophysiology}} - \field{volume}{5} - \field{year}{2011} - \endentry - \entry{GruenPhD}{book}{} - \name{author}{1}{}{% - {{hash=b50fe2ad55415c0cadde7f7184b65a0c}{Grün}{G\bibinitperiod}{S.}{S\bibinitperiod}{}{}{}{}}% - } - \list{publisher}{1}{% - {Verlag Harri Deutsch}% - } - \strng{namehash}{b50fe2ad55415c0cadde7f7184b65a0c} - \strng{fullhash}{b50fe2ad55415c0cadde7f7184b65a0c} - \field{sortinit}{G} - \field{sortinithash}{1c854ef9177a91bf894e66485bdbd3ed} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{note}{ISBN 3-8171-1506-7} - \field{series}{Reihe Physik, Band 60} - \field{title}{{Unitary Joint-Events in Multiple-Neuron Spiking Activity: Detection, Significance, and Interpretation}} - \field{year}{1996} - \endentry - \entry{Gruen02a}{article}{} - \name{author}{3}{}{% - {{hash=b50fe2ad55415c0cadde7f7184b65a0c}{Grün}{G\bibinitperiod}{S.}{S\bibinitperiod}{}{}{}{}}% - {{hash=54b69b31bebecc06a615a89151e20e56}{Diesmann}{D\bibinitperiod}{M.}{M\bibinitperiod}{}{}{}{}}% - {{hash=edc93e4bd0280a80f8f2d9b7e58bd8b3}{Aertsen}{A\bibinitperiod}{A.}{A\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{20875bf53ba00199f97bc089f54a7ff4} - \strng{fullhash}{20875bf53ba00199f97bc089f54a7ff4} - \field{sortinit}{G} - \field{sortinithash}{1c854ef9177a91bf894e66485bdbd3ed} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Neural Computation} - \field{title}{{`U}nitary {E}vents' in Multiple Single-Neuron Spiking Activity. {I.} {D}etection and Significance} - \field{volume}{14} - \field{year}{2002} - \field{pages}{43\bibrangedash 80} - \range{pages}{38} - \endentry - \entry{Gruen02b}{article}{} - \name{author}{3}{}{% - {{hash=b50fe2ad55415c0cadde7f7184b65a0c}{Grün}{G\bibinitperiod}{S.}{S\bibinitperiod}{}{}{}{}}% - {{hash=54b69b31bebecc06a615a89151e20e56}{Diesmann}{D\bibinitperiod}{M.}{M\bibinitperiod}{}{}{}{}}% - {{hash=edc93e4bd0280a80f8f2d9b7e58bd8b3}{Aertsen}{A\bibinitperiod}{A.}{A\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{20875bf53ba00199f97bc089f54a7ff4} - \strng{fullhash}{20875bf53ba00199f97bc089f54a7ff4} - \field{sortinit}{G} - \field{sortinithash}{1c854ef9177a91bf894e66485bdbd3ed} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Neural Computation} - \field{title}{{`U}nitary {E}vents' in Multiple Single-Neuron Spiking Activity. {II.} {N}on-{S}tationary Data} - \field{volume}{14} - \field{year}{2002} - \field{pages}{81\bibrangedash 119} - \range{pages}{39} - \endentry - \entry{Gruen03b}{article}{} - \name{author}{3}{}{% - {{hash=b50fe2ad55415c0cadde7f7184b65a0c}{Grün}{G\bibinitperiod}{S.}{S\bibinitperiod}{}{}{}{}}% - {{hash=5c57cf8680166d5eada49d3a930059f3}{Riehle}{R\bibinitperiod}{A.}{A\bibinitperiod}{}{}{}{}}% - {{hash=54b69b31bebecc06a615a89151e20e56}{Diesmann}{D\bibinitperiod}{M.}{M\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{fbf0b02892ca1ef3392a302c067d2c40} - \strng{fullhash}{fbf0b02892ca1ef3392a302c067d2c40} - \field{sortinit}{G} - \field{sortinithash}{1c854ef9177a91bf894e66485bdbd3ed} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Biological Cybernetics} - \field{title}{Effect of cross-trial nonstationarity on joint-spike events} - \field{volume}{88} - \field{year}{2003} - \field{pages}{335\bibrangedash 351} - \range{pages}{17} - \endentry - \entry{Gruen09}{article}{} - \name{author}{1}{}{% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{53e64085ba1c7e0fd687afbb264d9e30} - \strng{fullhash}{53e64085ba1c7e0fd687afbb264d9e30} - \field{sortinit}{G} - \field{sortinithash}{1c854ef9177a91bf894e66485bdbd3ed} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Journal of Neurophysiology} - \field{number}{101} - \field{title}{Data-driven significance estimation of precise spike correlation.} - \field{year}{2009} - \field{pages}{1126\bibrangedash 1140} - \range{pages}{15} - \endentry - \entry{GruenRotter10_Chap10}{inbook}{} - \name{author}{3}{}{% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - {{hash=a6e9c8b3b1dc422f6d4f5f42ab3051bc}{Diesmann}{D\bibinitperiod}{Markus}{M\bibinitperiod}{}{}{}{}}% - {{hash=83a9d134b8e1d6697bd710a1176471b1}{Aertsen}{A\bibinitperiod}{Ad}{A\bibinitperiod}{}{}{}{}}% - } - \name{editor}{2}{}{% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - {{hash=b2f53c7f6e46d970e3f44a698a690fc6}{Rotter}{R\bibinitperiod}{Stefan}{S\bibinitperiod}{}{}{}{}}% - } - \list{publisher}{1}{% - {Springer}% - } - \strng{namehash}{1debd6ae322071c9f68882a0d84560ef} - \strng{fullhash}{1debd6ae322071c9f68882a0d84560ef} - \field{sortinit}{G} - \field{sortinithash}{1c854ef9177a91bf894e66485bdbd3ed} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{chapter}{Unitary Event Analysis.} - \field{title}{Analysis of Parallel Spike Trains} - \field{year}{2010} - \endentry - \entry{Gruen99}{article}{} - \name{author}{5}{}{% - {{hash=b50fe2ad55415c0cadde7f7184b65a0c}{Grün}{G\bibinitperiod}{S.}{S\bibinitperiod}{}{}{}{}}% - {{hash=54b69b31bebecc06a615a89151e20e56}{Diesmann}{D\bibinitperiod}{M.}{M\bibinitperiod}{}{}{}{}}% - {{hash=4a7be839672bf6b5678a371baab5b069}{Grammont}{G\bibinitperiod}{F.}{F\bibinitperiod}{}{}{}{}}% - {{hash=5c57cf8680166d5eada49d3a930059f3}{Riehle}{R\bibinitperiod}{A.}{A\bibinitperiod}{}{}{}{}}% - {{hash=edc93e4bd0280a80f8f2d9b7e58bd8b3}{Aertsen}{A\bibinitperiod}{A.}{A\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{22540f06a27e5634301a23c42b7d1f9d} - \strng{fullhash}{f777839dbc56bc65b044c996c3918426} - \field{sortinit}{G} - \field{sortinithash}{1c854ef9177a91bf894e66485bdbd3ed} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Journal of Neuroscience Methods} - \field{title}{{Detecting unitary events without discretization of time}} - \field{volume}{94} - \field{year}{1999} - \field{pages}{67\bibrangedash 79} - \range{pages}{13} - \endentry - \entry{Ito11}{article}{} - \name{author}{4}{}{% - {{hash=14a981ebf618e2ffa6a59c2395e7ecee}{Ito}{I\bibinitperiod}{J}{J\bibinitperiod}{}{}{}{}}% - {{hash=1f6c6b83848bfc616c20ec6d8957f9a1}{Maldonado}{M\bibinitperiod}{P}{P\bibinitperiod}{}{}{}{}}% - {{hash=27515ed8ea87500fc8f5a164b4e2fcf9}{Singer}{S\bibinitperiod}{W}{W\bibinitperiod}{}{}{}{}}% - {{hash=d9099f36a962fb4744cdc95379e45704}{Grün}{G\bibinitperiod}{S}{S\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{e7b882946b826e2718b5f2e1d7391f85} - \strng{fullhash}{9931b1d3ec8901a8929b11b3c05b6af1} - \field{sortinit}{I} - \field{sortinithash}{25e99d37ba90f7c4fb20baf4e310faf3} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Cereb Cortex} - \field{title}{Saccade-related modulations of neuronal excitability support synchrony of visually elicited spikes} - \field{volume}{21} - \field{year}{2011} - \field{pages}{2482\bibrangedash 2497} - \range{pages}{16} - \endentry - \entry{Kilavik09}{article}{} - \name{author}{6}{}{% - {{hash=5e485522369879dd6d2217d47b28e1ad}{Kilavik}{K\bibinitperiod}{B.\bibnamedelimi E.}{B\bibinitperiod\bibinitdelim E\bibinitperiod}{}{}{}{}}% - {{hash=965dfaaeb5bda9a53305156ff3332c45}{Roux}{R\bibinitperiod}{S.}{S\bibinitperiod}{}{}{}{}}% - {{hash=49950437c54d71df53dc19a89c069afb}{Ponce-Alvarez}{P\bibinithyphendelim A\bibinitperiod}{A.}{A\bibinitperiod}{}{}{}{}}% - {{hash=cb48f0b5b58b913e666b64b93eb21721}{Confais}{C\bibinitperiod}{J.}{J\bibinitperiod}{}{}{}{}}% - {{hash=b50fe2ad55415c0cadde7f7184b65a0c}{Grün}{G\bibinitperiod}{S.}{S\bibinitperiod}{}{}{}{}}% - {{hash=5c57cf8680166d5eada49d3a930059f3}{Riehle}{R\bibinitperiod}{A.}{A\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{29346e996225efad73d5e033aaee0f93} - \strng{fullhash}{d0d3e2e7684bddcb2b963552be5e17a2} - \field{sortinit}{K} - \field{sortinithash}{a7d5b3aec5a0890aae7baf85a209abfc} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Journal of Neuroscience} - \field{title}{Long-term Modifications in Motor Cortical Dynamics induced by intensive practice} - \field{volume}{29} - \field{year}{2009} - \field{pages}{12653\bibrangedash 12663} - \range{pages}{11} - \endentry - \entry{Louis10}{article}{} - \name{author}{4}{}{% - {{hash=165807253efb456afc14f5c5a637e435}{Louis}{L\bibinitperiod}{S.}{S\bibinitperiod}{}{}{}{}}% - {{hash=3990c3843f0ee4bf5411e15565b676ea}{Gerstein}{G\bibinitperiod}{G.\bibnamedelimi L.}{G\bibinitperiod\bibinitdelim L\bibinitperiod}{}{}{}{}}% - {{hash=b50fe2ad55415c0cadde7f7184b65a0c}{Grün}{G\bibinitperiod}{S.}{S\bibinitperiod}{}{}{}{}}% - {{hash=54b69b31bebecc06a615a89151e20e56}{Diesmann}{D\bibinitperiod}{M.}{M\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{dc763cec9330cb69bcd81e0acc5ebb74} - \strng{fullhash}{f069756d45eaeb49053fdbe99ce68ea0} - \field{sortinit}{L} - \field{sortinithash}{872351f18d0f736066eda0bf18bfa4f7} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Front. Comput. Neurosci.} - \field{title}{Surrogate spike train generation through dithering in operational time} - \field{volume}{4} - \field{year}{2010} - \field{pages}{127} - \range{pages}{1} - \endentry - \entry{Maldonado08}{article}{} - \name{author}{6}{}{% - {{hash=73e46ed2e548d51f8bbc68b2929767e2}{Maldonado}{M\bibinitperiod}{Pedro}{P\bibinitperiod}{}{}{}{}}% - {{hash=fba9fec209f577e1c72e661febb46406}{Babul}{B\bibinitperiod}{Cecilia}{C\bibinitperiod}{}{}{}{}}% - {{hash=974000371a3ba2b0a655dec8d315a71a}{Singer}{S\bibinitperiod}{Wolf}{W\bibinitperiod}{}{}{}{}}% - {{hash=e45fbcfce0affb5e78e238a2d024f924}{Rodriguez}{R\bibinitperiod}{Eugenio}{E\bibinitperiod}{}{}{}{}}% - {{hash=7c609bc456fd4768d19eaf6872cbad0f}{Berger}{B\bibinitperiod}{Denise}{D\bibinitperiod}{}{}{}{}}% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{6b639187cc4cad4a12a11b8d23e7117a} - \strng{fullhash}{77ceef782830c7c29c8c133d84f4244b} - \field{sortinit}{M} - \field{sortinithash}{2684bec41e9697b92699b46491061da2} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Journal of Neurophysiology} - \field{title}{Synchronization of Neuronal Responses in Primary Visual Cortex of Monkeys Viewing Natural Images} - \field{volume}{100} - \field{year}{2008} - \field{pages}{1523\bibrangedash 1532} - \range{pages}{10} - \endentry - \entry{Nakahara02}{article}{} - \name{author}{2}{}{% - {{hash=d4691036508f4bf83c916efd67be3f6a}{Nakahara}{N\bibinitperiod}{H.}{H\bibinitperiod}{}{}{}{}}% - {{hash=091eb0563ff135bac3f03117f00ca92e}{Amari}{A\bibinitperiod}{S.}{S\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{beda892cbfdb1ea78b555798b612f2ad} - \strng{fullhash}{beda892cbfdb1ea78b555798b612f2ad} - \field{sortinit}{N} - \field{sortinithash}{925374ca63e7594de7fafdb83e64d41d} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Neural Computation} - \field{title}{Information-geometric measure for neural spikes} - \field{volume}{14} - \field{year}{2002} - \field{pages}{2269\bibrangedash 2316} - \range{pages}{48} - \endentry - \entry{Pipa2013}{article}{} - \name{author}{3}{}{% - {{hash=fc589f8d0d85e42c9326684c4e155520}{Pipa}{P\bibinitperiod}{Gordon}{G\bibinitperiod}{}{}{}{}}% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - {{hash=e4a1fa8bad8c1a509d287620b07ebbdf}{Vreeswijk}{V\bibinitperiod}{Carl\bibnamedelima Van}{C\bibinitperiod\bibinitdelim V\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{2e0752b048bae173b25fd34b27aee9c9} - \strng{fullhash}{2e0752b048bae173b25fd34b27aee9c9} - \field{sortinit}{P} - \field{sortinithash}{c0a4896d0e424f9ca4d7f14f2b3428e7} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Neural Computation} - \field{title}{{Impact of Spike Train Autostructure on Probability Distribution of Joint Spike Events}} - \field{volume}{1163} - \field{year}{2013} - \field{pages}{1123\bibrangedash 1163} - \range{pages}{41} - \endentry - \entry{Riehle2000}{article}{} - \name{author}{4}{}{% - {{hash=0840237d57f02a400049f66f51b385d9}{Riehle}{R\bibinitperiod}{Alexa}{A\bibinitperiod}{}{}{}{}}% - {{hash=6b07390586ee048fffae07a3e2dcee7c}{Grammont}{G\bibinitperiod}{Franck}{F\bibinitperiod}{}{}{}{}}% - {{hash=a6e9c8b3b1dc422f6d4f5f42ab3051bc}{Diesmann}{D\bibinitperiod}{Markus}{M\bibinitperiod}{}{}{}{}}% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{88d15238933e9482c8be9dbf1e0af060} - \strng{fullhash}{a618f00c8d1da33451510013728150f3} - \field{sortinit}{R} - \field{sortinithash}{c7387613477035a752d935acfc3e3ea2} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Journal of Physiology Paris} - \field{title}{Dynamical changes and temporal precision of synchronized spiking activity in monkey motor cortex during movement preparation} - \field{volume}{94} - \field{year}{2000} - \field{pages}{569\bibrangedash 582} - \range{pages}{14} - \endentry - \entry{Riehle97}{article}{} - \name{author}{4}{}{% - {{hash=0840237d57f02a400049f66f51b385d9}{Riehle}{R\bibinitperiod}{Alexa}{A\bibinitperiod}{}{}{}{}}% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - {{hash=a6e9c8b3b1dc422f6d4f5f42ab3051bc}{Diesmann}{D\bibinitperiod}{Markus}{M\bibinitperiod}{}{}{}{}}% - {{hash=83a9d134b8e1d6697bd710a1176471b1}{Aertsen}{A\bibinitperiod}{Ad}{A\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{88d15238933e9482c8be9dbf1e0af060} - \strng{fullhash}{2cc98b0e4bf7a8003861316b298c5392} - \field{sortinit}{R} - \field{sortinithash}{c7387613477035a752d935acfc3e3ea2} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Science} - \field{title}{Spike Synchronization and Rate Modulation Differentially Involved in Motor Cortical Function} - \field{volume}{278} - \field{year}{1997} - \field{pages}{1950\bibrangedash 1953} - \range{pages}{4} - \endentry - \entry{Pullrequest_UE}{article}{} - \name{author}{1}{}{% - {{hash=f49c65cd533b79d8aa1bba32d8cf88b0}{Rostami}{R\bibinitperiod}{Vahid}{V\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{f49c65cd533b79d8aa1bba32d8cf88b0} - \strng{fullhash}{f49c65cd533b79d8aa1bba32d8cf88b0} - \field{sortinit}{R} - \field{sortinithash}{c7387613477035a752d935acfc3e3ea2} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Elephant Github} - \field{title}{Pull Request of the Unitary Events method in Elephant} - \field{year}{2016} - \verb{url} - \verb https://github.com/NeuralEnsemble/elephant/pull/64 - \endverb - \endentry - \entry{Zehl2016}{article}{} - \name{author}{10}{}{% - {{hash=86a847fed233df0376116790ede82340}{Zehl}{Z\bibinitperiod}{Lyuba}{L\bibinitperiod}{}{}{}{}}% - {{hash=b124ef01d8cff212ad2f689dc37ff92d}{Jaillet}{J\bibinitperiod}{Florent}{F\bibinitperiod}{}{}{}{}}% - {{hash=7372b5a0a87df8dec679ca9af13decde}{Stoewer}{S\bibinitperiod}{Adrian}{A\bibinitperiod}{}{}{}{}}% - {{hash=23e27f78671208ddad6a55ffec7d1cce}{Grewe}{G\bibinitperiod}{Jan}{J\bibinitperiod}{}{}{}{}}% - {{hash=5c54dce88d4e8d05a24fda7d6f7c898e}{Sobolev}{S\bibinitperiod}{Andrey}{A\bibinitperiod}{}{}{}{}}% - {{hash=739fea8110d6909bf6645f56b5fc629c}{Wachtler}{W\bibinitperiod}{Thomas}{T\bibinitperiod}{}{}{}{}}% - {{hash=2ee59795ddc280f865e75cc6caba1bbf}{Brochier}{B\bibinitperiod}{Thomas\bibnamedelima G.}{T\bibinitperiod\bibinitdelim G\bibinitperiod}{}{}{}{}}% - {{hash=0840237d57f02a400049f66f51b385d9}{Riehle}{R\bibinitperiod}{Alexa}{A\bibinitperiod}{}{}{}{}}% - {{hash=f6f72f78b262d4fcff83ad67927cd29c}{Denker}{D\bibinitperiod}{Michael}{M\bibinitperiod}{}{}{}{}}% - {{hash=53e64085ba1c7e0fd687afbb264d9e30}{Grün}{G\bibinitperiod}{Sonja}{S\bibinitperiod}{}{}{}{}}% - } - \strng{namehash}{7e24e9bf7f42aa0da5431c48bcf7bbac} - \strng{fullhash}{916dabca2db4465fdd5d0358d6fc71d8} - \field{sortinit}{Z} - \field{sortinithash}{fdda4caaa6b5fa63e0c081dcb159543a} - \field{labelnamesource}{author} - \field{labeltitlesource}{title} - \field{journaltitle}{Frontiers in Neuroinformatics} - \field{title}{{Handling Metadata in a Neurophysiology Laboratory}} - \field{volume}{10} - \field{year}{2016} - \field{pages}{26} - \range{pages}{1} - \endentry - \endsortlist -\endrefsection -\endinput - diff --git a/article/VRostami_JIto_MDenker_SGruen_2016.blg b/article/VRostami_JIto_MDenker_SGruen_2016.blg deleted file mode 100644 index 500093d..0000000 --- a/article/VRostami_JIto_MDenker_SGruen_2016.blg +++ /dev/null @@ -1,15 +0,0 @@ -[0] Config.pm:327> INFO - This is Biber 2.2 -[0] Config.pm:330> INFO - Logfile is 'VRostami_JIto_MDenker_SGruen_2016.blg' -[44] biber:286> INFO - === So Apr 23, 2017, 15:53:19 -[44] Biber.pm:334> INFO - Reading 'VRostami_JIto_MDenker_SGruen_2016.bcf' -[146] Biber.pm:677> INFO - Found 28 citekeys in bib section 0 -[164] Biber.pm:3054> INFO - Processing section 0 -[179] Biber.pm:3193> INFO - Looking for bibtex format file 'VRostami_JIto_MDenker_SGruen_2016.bib' for section 0 -[181] bibtex.pm:1072> INFO - Decoding LaTeX character macros into UTF-8 -[184] bibtex.pm:935> INFO - Found BibTeX data source 'VRostami_JIto_MDenker_SGruen_2016.bib' -[532] UCollate.pm:66> INFO - Overriding locale 'en-US' defaults 'variable = shifted' with 'variable = non-ignorable' -[533] UCollate.pm:66> INFO - Overriding locale 'en-US' defaults 'normalization = NFD' with 'normalization = prenormalized' -[533] Biber.pm:2931> INFO - Sorting list 'nty' of type 'entry' with scheme 'nty' and locale 'en-US' -[533] Biber.pm:2935> INFO - No sort tailoring available for locale 'en-US' -[594] bbl.pm:505> INFO - Writing 'VRostami_JIto_MDenker_SGruen_2016.bbl' with encoding 'UTF-8' -[598] bbl.pm:596> INFO - Output to VRostami_JIto_MDenker_SGruen_2016.bbl diff --git a/article/VRostami_JIto_MDenker_SGruen_2016.pdf b/article/VRostami_JIto_MDenker_SGruen_2016.pdf deleted file mode 100644 index 5605178..0000000 Binary files a/article/VRostami_JIto_MDenker_SGruen_2016.pdf and /dev/null differ diff --git a/article/VRostami_JIto_MDenker_SGruen_2016.run.xml b/article/VRostami_JIto_MDenker_SGruen_2016.run.xml deleted file mode 100644 index 61e4094..0000000 --- a/article/VRostami_JIto_MDenker_SGruen_2016.run.xml +++ /dev/null @@ -1,85 +0,0 @@ - - - - - - - - - - - - - - - - - - - - - - - - -]> - - - latex - - VRostami_JIto_MDenker_SGruen_2016.bcf - - - VRostami_JIto_MDenker_SGruen_2016.bbl - - - blx-dm.def - blx-compat.def - biblatex.def - numeric.bbx - standard.bbx - numeric.cbx - biblatex.cfg - english.lbx - - - - biber - - biber - VRostami_JIto_MDenker_SGruen_2016 - - - VRostami_JIto_MDenker_SGruen_2016.bcf - - - VRostami_JIto_MDenker_SGruen_2016.bbl - - - VRostami_JIto_MDenker_SGruen_2016.bbl - - - VRostami_JIto_MDenker_SGruen_2016.bcf - - - VRostami_JIto_MDenker_SGruen_2016.bib - - - diff --git a/article/VRostami_JIto_MDenker_SGruen_2016.tex b/article/VRostami_JIto_MDenker_SGruen_2016.tex deleted file mode 100644 index f3d4bfb..0000000 --- a/article/VRostami_JIto_MDenker_SGruen_2016.tex +++ /dev/null @@ -1,992 +0,0 @@ -\documentclass[10pt,a4paper,onecolumn]{article} -% \usepackage[utf8]{inputenc} -\usepackage{marginnote} -\usepackage{graphicx} -\usepackage{xcolor} -\usepackage{authblk,etoolbox} -\usepackage{titlesec} -\usepackage{calc} -\usepackage{hyperref} -\hypersetup{breaklinks=true, - bookmarks=true, - pdfauthor= -{ - Vahid Rostami, - Junji Ito, - Michael Denker, - Sonja Grün, - }, - pdftitle= -{ -[Re] Spike Synchronization and Rate Modulation Differentially Involved in -Motor Cortical Function -}, - colorlinks=true, - citecolor=blue, - urlcolor=blue, - linkcolor=blue, - pdfborder={0 0 0}} -\urlstyle{same} -\usepackage{tcolorbox} -\usepackage{ragged2e} -\usepackage{fontspec} -\usepackage{fontawesome} -\usepackage{caption} -\usepackage{listings} -\lstnewenvironment{code}{\lstset{language=Haskell,basicstyle=\small\ttfamily}}{} - - - -%\usepackage{fancyvrb} -%\VerbatimFootnotes -%\usepackage{graphicx} -%\usepackage{mdframed} -%\newmdenv[backgroundcolor=lightgray]{Shaded} - - -\usepackage{longtable,booktabs} - -\usepackage[ - backend=biber, -% style=alphabetic, -% citestyle=numeric -]{biblatex} -\bibliography{VRostami_JIto_MDenker_SGruen_2016.bib} - - - -% --- Macros ------------------------------------------------------------------ -\renewcommand*{\bibfont}{\small \sffamily} - -\definecolor{red}{HTML}{CF232B} -\newcommand{\ReScience}{Re{\bfseries \textcolor{red}{Science}}} - -\newtcolorbox{rebox} - {colback=blue!5!white, colframe=blue!40!white, - boxrule=0.5pt, arc=2pt, fonttitle=\sffamily\scshape\bfseries, - left=6pt, right=20pt, top=6pt, bottom=6pt} - -\newtcolorbox{repobox} - {colback=red, colframe=red!75!black, - boxrule=0.5pt, arc=2pt, left=6pt, right=6pt, top=3pt, bottom=3pt} - -% fix for pandoc 1.14 -\newcommand{\tightlist}{% - \setlength{\itemsep}{1pt}\setlength{\parskip}{0pt}\setlength{\parsep}{0pt}} - -% --- Style ------------------------------------------------------------------- -\renewcommand*{\bibfont}{\small \sffamily} -\renewcommand{\captionfont}{\small\sffamily} -\renewcommand{\captionlabelfont}{\bfseries} - -\makeatletter -\renewcommand\@biblabel[1]{{\bf #1.}} -\makeatother - -% --- Page layout ------------------------------------------------------------- -\usepackage[top=3.5cm, bottom=3cm, right=1.5cm, left=1.5cm, - headheight=2.2cm, reversemp, includemp, marginparwidth=4.5cm]{geometry} - -% --- Section/SubSection/SubSubSection ---------------------------------------- -\titleformat{\section} - {\normalfont\sffamily\Large\bfseries} - {}{0pt}{} -\titleformat{\subsection} - {\normalfont\sffamily\large\bfseries} - {}{0pt}{} -\titleformat{\subsubsection} - {\normalfont\sffamily\bfseries} - {}{0pt}{} -\titleformat*{\paragraph} - {\sffamily\normalsize} - - -% --- Header / Footer --------------------------------------------------------- -\usepackage{fancyhdr} -\pagestyle{fancy} -%\renewcommand{\headrulewidth}{0.50pt} -\renewcommand{\headrulewidth}{0pt} -\fancyhead[L]{\hspace{-1cm}\includegraphics[width=4.0cm]{rescience-logo.pdf}} -\fancyhead[C]{} -\fancyhead[R]{} -\renewcommand{\footrulewidth}{0.25pt} - -\fancyfoot[L]{\hypersetup{urlcolor=red} - \sffamily \ReScience~$\vert$ - \href{http://rescience.github.io}{rescience.github.io} - \hypersetup{urlcolor=blue}} -\fancyfoot[C]{\sffamily \thepage} -\fancyfoot[R]{\sffamily Nov 2016 $\vert$ - Volume \textbf{1} $\vert$ - Issue \textbf{1}} -\pagestyle{fancy} -\makeatletter -\let\ps@plain\ps@fancy -\fancyheadoffset[L]{4.5cm} -\fancyfootoffset[L]{4.5cm} - -% --- Title / Authors --------------------------------------------------------- -% patch \maketitle so that it doesn't center -\patchcmd{\@maketitle}{center}{flushleft}{}{} -\patchcmd{\@maketitle}{center}{flushleft}{}{} -% patch \maketitle so that the font size for the title is normal -\patchcmd{\@maketitle}{\LARGE}{\LARGE\sffamily}{}{} -% patch the patch by authblk so that the author block is flush left -\def\maketitle{{% - \renewenvironment{tabular}[2][] - {\begin{flushleft}} - {\end{flushleft}} - \AB@maketitle}} -\makeatletter -\renewcommand\AB@affilsepx{ \protect\Affilfont} -%\renewcommand\AB@affilnote[1]{{\bfseries #1}\hspace{2pt}} -\renewcommand\AB@affilnote[1]{{\bfseries #1}\hspace{3pt}} -\makeatother -\renewcommand\Authfont{\sffamily\bfseries} -\renewcommand\Affilfont{\sffamily\small\mdseries} -\setlength{\affilsep}{1em} - -\LetLtxMacro{\OldIncludegraphics}{\includegraphics} -\renewcommand{\includegraphics}[2][]{\OldIncludegraphics[width=12cm, #1]{#2}} - - -% --- Document ---------------------------------------------------------------- -\title{[Re] Spike Synchronization and Rate Modulation Differentially Involved in -Motor Cortical Function} - - \usepackage{authblk} - \author[1]{Vahid Rostami} - \author[1]{Junji Ito} - \author[1]{Michael Denker} - \author[1,2]{Sonja Grün} - \affil[1]{Inst. of Neuroscience \& Medicine (INM-6) and Inst. for Advanced -Simulation (IAS-6), JARA Brain Institute I, Jülich Research Center, -Jülich, Germany} - \affil[2]{Theoretical Systems Neurobiology, RWTH Aachen University, Aachen, -Germany} - -\date{\vspace{-5mm} - \sffamily \small \href{mailto:v.rostami@fz-juelich.de}{v.rostami@fz-juelich.de}} - - -\setlength\LTleft{0pt} -\setlength\LTright{0pt} - - -\begin{document} -\maketitle - -\marginpar{ - %\hrule - \sffamily\small - %\vspace{2mm} - {\bfseries Editor}\\ - Name Surname\\ - - {\bfseries Reviewers}\\ - Name Surname\\ - Name Surname\\ - - {\bfseries Received} Nov, 1, 2016\\ - {\bfseries Accepted} Nov, 1, 2016\\ - {\bfseries Published} Nov, 1, 2016\\ - - {\bfseries Licence} \href{http://creativecommons.org/licenses/by/4.0/}{CC-BY} - - \begin{flushleft} - {\bfseries Competing Interests:}\\ - The authors have declared that no competing interests exist. - \end{flushleft} - - \hrule - \vspace{3mm} - - \hypersetup{urlcolor=white} - - \vspace{-1mm} - \begin{repobox} - \bfseries\normalsize - \href{http://github.com/rescience/rescience-submission/article}{\faGithubAlt~Article repository} - \end{repobox} - \vspace{-1mm} - \begin{repobox} - \bfseries\normalsize - \href{http://github.com/rescience/rescience-submission/code}{\faGithubAlt~Code repository} - \end{repobox} - \hypersetup{urlcolor=blue} -} - -\begin{rebox} -\sffamily {\bfseries A reference implementation of} -\small -\begin{flushleft} -\begin{itemize} - \item[→] Spike synchronization and rate modulation differentially involved in -motor cortical function. Alexa Riehle, Sonja Grün, Markus Diesmann, and -Ad Aertsen (1997) Science 278:1950-1953. DOI:10.1126/science.278.5345.19 -50 - \end{itemize}\par -\end{flushleft} -\end{rebox} - - -\section{Introduction}\label{introduction} - -Understanding how information is processed by networks of neurons in the -brain is a major goal in neuroscience. There has been a long-standing -debate in the community on the contribution of the firing activity of -individual neurons or populations of neurons to information processing: -one perspective is that the firing rate of the neurons, i.e.~the number -of spikes per time unit, represents information or is relevant for the -processing. Another perspective is that neurons communicate through -precisely timed coordination of spiking, a view that results from the -insight that a neuron fires most efficiently if it receives synchronous -spike input, i.e., in the range of a few milliseconds -\autocite{Abeles82}. - -To test whether temporal coordination of spiking activity is indeed -relevant for neuronal information processing, advanced data analysis -methods are required that perform correlation analysis between -simultaneously recorded single unit spike trains. The Unitary Events -(UE) analysis method \autocites{GruenPhD}{Gruen99}{Gruen02a}{Gruen02b} -is able to extract significant spike synchrony between neuronal -activities that is beyond what is expected by chance (given the rates) -and to follow the dynamics of this coordination. The tricky issue with -such analyses is that one has to take into account that experimental -spike data are typically non-stationary in the sense that the firing -rates are not stationary in time and not homogeneous across trials, and -that the statistics of the individual spike trains deviate from those of -a Poisson process \autocites{Gruen09}{GruenRotter10_Chap10}. If such -features are ignored there is a considerable danger of occurrence of -false positives and thus wrong interpretation of the data. In -particular, changes in the firing rates are the most prominent -generators of false positives if ignored. The original UE method -considers this aspect by performing the analysis in a sliding time -window fashion. In later versions of the analysis method other features -were corrected for by considering the respective features in the -null-hypothesis of the tests -\autocites{Gruen03b}{Maldonado08}{Louis10}{Pipa2013}, either by using -extended analytical descriptions of the null-hypotheses or by use of -surrogate methods \autocites{Gruen09}{GruenRotter10_Chap10}{Louis10}. - -The UE method had been applied to experimental parallel spike data from, -e.g., the motor cortex of awake behaving non-human primates -\autocites{Riehle97}{Gruen99}{Grammont99}{Riehle2000}{Gruen03b}{Kilavik09}{Denker10}{Denker11} -and to data from visual cortices \autocites{Maldonado08}{Ito11}. -Generally it was found that UEs, i.e., synchronous spike events across -neurons that are in excess of the expectation, occur in relation to -behavior, e.g., when the animal expects a signal to occur, however the -signal does not occur \autocite{Riehle97}. This finding, in particular, -demonstrates the core feature of the method: by performing a time -resolved analysis, the method accounts for changes in the firing rates -and captures modulations of significant spike synchrony in time. It was -also shown that the time of occurrence of UEs may change to a new -requested timing in the behavior during a learning process -\autocite{Kilavik09}. - -Several of these studies based on the UE method, in particular -\autocite{Riehle97}, have been widely cited and the method has been -recognized as one of the standard tools to analyze temporal coordination -of neuronal spiking activities \autocites{Brown04_456}{Nakahara02}. The -analysis method has been and is taught in international data analysis -courses. However, a publicly available and open source implementation of -the UE method had not been available. Only very recently, one of the -authors of this study (VR) reimplemented the UE analysis as part of the -Electrophysiology Analysis Toolbox\footnote{\url{http://neuralensemble.org/elephant/}} -(Elephant), a Python library that provides implementations for the -analysis of electrophysiological data. This reimplementation of the -method was particularly required for, in addition to the general -motivation for a reimplementation to confirm the results not depending -on small implementation details or mistakes, the following two specific -reasons. - -First, the custom data object model used to represent the primary data -and metadata in the original analysis code was not documented. -Therefore, any data represented in a specific file format had to be -converted by implementing a custom data loading routine for this data -model. Our new implementation of the UE method is part of the Elephant -and Neo\footnote{\url{http://neuralensemble.org/neo/}} libraries and is -based on the internal data object model provided by the Neo library -\autocite{Garcia14}, a package for representing electrophysiology data -in Python. Neo provides support for reading a wide range of -neurophysiology proprietary file formats, and supports writing to a -subset of these formats, including non-proprietary formats such as HDF5. - -Second, the implementation of the UE method in the original publication -has experienced several updates after its publication in order to -include improvements and extensions of the method to accommodate more -features of experimental data. Since no version control was employed in -this development process, the original code used in Riehle et al. -\autocite{Riehle97} was lost at some point. Considering that no -systematic testing was performed each time the code base of the Unitary -Event analysis was updated after the original publication to check -whether the code gives the same results as before, it was not clear -whether the latest version could exactly reproduce the results of -\autocite{Riehle97}. - -In this paper we illustrate the successful reproduction of the results -shown in \autocite{Riehle97} using our new Python implementation of the -UE method. In particular, we reproduce Figure 2 and Figure 4A of the -original paper, which represent the central results of the original -study. The remainder of the original paper consisted of more example -analyses of individual data sets and a meta statistics across many data -sets, all of which were based on the same analysis method and thus do -not provide additional insight when reproduced. In the original -publication the authors used two different UE implementations: one -implemented by one of the coauthors of the present study (SG) in -IDL\footnote{\url{http://www.harrisgeospatial.com/ProductsandSolutions/GeospatialProducts/IDL.aspx}}, -and the other in Matlab (Mathworks, Nattick, MA) implemented later by -Markus Diesmann (MD) and SG. The IDL implementation is not available -anymore. At a later point we were provided with a Matlab implementation -of the extended UE method, however it does not preserve the original -implementation used in \autocite{Riehle97} and was not considered in -this study. - -For the reproduction of the original results we contacted and -communicated with Alexa Riehle (AR), CNRS-AMU, Marseille, and MD, -Research Centre Jülich. AR is the first author of the original -publication and performed the experiments and data analysis. MD is the -third author of the original publication and contributed with the Matlab -implementation and quality checks of the software implementations. The -reproduction of the results of \autocite{Riehle97} would not have been -possible without contacting these authors, since the information in the -original publication is not sufficient for reproducing the results. AR -provided us with the original data and information, including an old -written report by MD, which was crucial to reproduce Figure 4A. - -\section{Methods}\label{methods} - -For reproducing the results of \autocite{Riehle97} we use our -reimplementation of the UE analysis method in Python which is made -available in the \texttt{unitary\_event\_analysis} module of the -Elephant library (accepted pull-request: \autocite{Pullrequest_UE}). The -method is accompanied by unit tests for individual functions (test -coverage: \(88.54\%\)) and documented as part of the library -documentation. The structure and algorithm of our UE implementation is -explained in the pseudo-code shown below. - -In the following, we will explain the algorithm in detail. The primary -data entering the UE method is a set of parallel, i.e., simultaneously -recorded, spike trains, recorded in one or multiple trials. In Elephant, -an individual spike train of a particular neuron in a particular trial -data is represented as a \texttt{Spiketrain} object in the data object -model provided by the Neo library, which stores the time points of spike -occurrences along with additional information describing the spike -train, such as the start and end times. The UE method consumes a nested -list of \texttt{Spiketrain} objects relating to the spike trains of -individual neurons in the individual trials. In order to perform the -necessary calculations, the spike data must be converted to an -alternative time-binned representation (line 1 in the pseudo code) where -the parallel spike trains are stored as binary sequences of ones -(marking time bins containing at least 1 spike) and zeros (bins with no -spike). The bin size is a parameter provided as input (parameter -\emph{\texttt{bin\_size}} in the pseudo code) to the analysis, and -defines the temporal precision in detecting spike synchrony. A spike -pattern, which is a representation of spike synchrony in the analysis, -is defined as a specific vector of zeros and ones in one time bin of a -given trial across all neurons. Since the aim of the UE method is to -detect significant spike synchrony, the total number of spike patterns -of concern is given by \(2^{N}-N-1\), where \(N\) is the number of -neurons, the first term is the number of all possible spike patterns, -the second term is the number of spike patterns composed of only one -spike, and the third term is the number of the pattern without a spike. -In order to limit the analysis to a set of patterns of interest, the -input parameter \emph{\texttt{pattern\_hash\_values}} specifies the -patterns to consider in the analysis in the form of a hash value which -uniquely represents each spike pattern (line 2 in the pseudo code). The -hash value is obtained by interpreting the binary spike pattern as a -binary number, where the \(n\)-th neuron is represented by bit \(n-1\). - -The UE analysis is performed in a sliding time window fashion, i.e.~the -data in each window are analyzed separately. This approach is chosen to -account for potential changes of the firing rates in time and to follow -the dynamics of the correlation. Here, a sliding time window is defined -based on trial time, meaning that a certain window position includes the -activity of all neurons in all trials in a certain time interval of the -trial. In the algorithm, at each position of the window, the data -contained in the window are extracted in order to compute the -significance of the specified patterns (line 3 in the pseudo code). For -doing that the UE analysis requires the empirical number of occurrences -of each pattern as well as its expected number given the firing rates. -While the empirical number is directly extracted from the data (line 5 -in the pseudo code), the method to compute the expected number can be -chosen using the input parameter \emph{\texttt{method}} depending on the -assumptions regarding the data (line 6-19 in the pseudo code). For -cross-trial homogeneous data, selecting -\emph{``analytic\_TrialAverage''} as \emph{\texttt{method}} (line 7 in -the pseudo code) computes the expected number by estimating the rate of -each neuron from the average spike count across trials -\autocites{Gruen02a}{Gruen02b}. For data with\\ -\includegraphics{UE_algorithm.eps}\\ -cross-trial variability, the \emph{``analytic\_TrialByTrial''} method -(line 10 in the pseudo code) should be used, which accounts for -cross-trial changes in the firing rates by computing the expected number -of occurrences of the spike pattern based on the product of single-trial -estimates of firing rates \autocite{Gruen03b}. Both methods perform a -parametric test, where the number of occurrences of the spike pattern is -assumed to be a Poisson-distributed. As an alternative non-parametric -option, selecting \emph{``surrogate\_TrialByTrial''} as -\emph{\texttt{method}} (line 15 in the pseudo code) will numerically -compute the distribution of the expected number by implementing the -null-hypothesis based on surrogate spike trains \autocite{Gruen09}. In -the following, we will explain these methods in greater detail. - -In case of selecting \emph{``analytic\_TrialAverage''} - as used in the -remainder of this study for the reproduction of \autocite{Riehle97} - -the number of spikes per neuron within the sliding window is summed -across trials and divided by the number of trials and bins contained, -thus yielding the average probability \(p_{i}\) to have a spike of -neuron \(i\) in a bin of the time window. The probability to find a -particular pattern by chance in a bin is then computed by multiplication -of the relevant probabilities, e.g.~for a pattern {[}1,0,1{]} the -probability \(p\) of occurrence is given by -\(p_{[1,0,1]}=p_{1}*(1-p_{2})*p_{3}\). Note that (\(1-p_{2}\)) is the -probability for neuron 2 to contribute no spike to the pattern. The -expected number of pattern occurrences, computed as the product of the -occurrence probability \(p\) of the pattern, e.g. \(p_{[1,0,1]}\), and -the number of bins (across all trials) covered by the sliding window. -The distribution of pattern occurrence numbers is given by a Poisson -distribution with the mean equal to the expected number. - -Alternatively, for the \emph{``analytic\_TrialByTrial''} method, the -firing probability of each neuron is calculated in a trial-by-trial -manner based on the spike counts per trial. The probability \(p\) of -finding the pattern by chance is calculated by summing the products of -the firing probabilities obtained individually from each trial. As for -the trial-averaging method, the expected number of pattern occurrences -is given by multiplication with the number of bins of the trial in the -sliding window, used as the mean of a Poisson distribution to obtain the -distribution of expected pattern occurrence numbers. - -As a third alternative, a surrogate method for estimating the expected -number can be selected using \emph{``surrogate\_TrialByTrial''} for the -parameter \emph{\texttt{method}}. In this Monte-Carlo approach, a -surrogate version of the spike trains is generated repeatedly, and from -each surrogate the number of occurrences of the pattern of interest is -counted. The method by which surrogates are generated from the input -spike trains is spike time randomization of the spikes per trial and per -neuron within the sliding window. The pattern counts obtained from this -procedure form a distribution of the expected number of occurrences of -the pattern, thus implementing the null-hypothesis under the constraints -implied by the surrogate method. - -The distribution obtained by either of the three methods above is then -used for the significance test of the pattern on the basis of the -empirical occurrence count. The p-value resulting from the test is then -transformed by a logarithmic transformation to the surprise value (line -20 in the pseudo code), which indicates by positive or negative values -more or less occurrences of the pattern than expected by chance, -respectively. If the p-value is below a fixed prescribed level -(e.g.~below \(5\%\), which corresponds to a surprise value exceeding -\(1.27\)), the occurrences of the spike pattern under investigation in -the sliding window are marked as UEs for that pattern. This procedure is -performed for each pattern of interest, and in each sliding window. - -In the present study, in order to reproduce the original results we used -the \emph{``analytic\_TrialAverage''} method, which reflects the -analysis performed in the original publication \autocite{Riehle97}. The -\emph{``analytic\_TrialByTrial''} and \emph{``surrogate\_TrialByTrial''} -methods are extensions of the original UE method, which were developed -after the original publication and introduced in subsequent works -\autocites{Gruen03b}{Gruen09}. - -Our reimplementation of the UE method is based on the data object model -provided by the Neo library, upon which the Elephant library is based. -The Neo library provides loading routines for a variety of data formats, -including proprietary and generic data formats. The data sets available -for reproducing Figures 2 and 4A of \autocite{Riehle97} were -tab-separated ASCII text files containing two columns of integers -(informally often referred to as ``GDF-format''): the first column -provides event codes (behavioral events or neuron IDs), and the second -column contains the time of the occurrence of these events (time -stamps). The units of the time stamps are not contained in the data -file. We partly extracted metadata information, in particular the time -units and the meaning of the event codes, from a Matlab routine -(provided by AR) operating on the GDF data file. However, only after -further communication with AR we were able to identify the exact meaning -of the content of the data files. Using this information, we wrote a new -loading routine that loads the GDF data as Neo data objects. - -Our reimplementation uses the \texttt{conversion} module of Elephant for -converting the spike data (represented as a series of timestamps) into -the binary sequence to guarantee a unique, global binning mechanism for -all analysis methods provided in Elephant. The bin size to be set for -the analysis was extracted from the original publication. However, -defining the time point to start the binning of each single trial data -required to know the alignment event in each trial and how much time -before this event (pre-time) is considered. Since this information was -not documented in the original paper, we tried several possibilities -until we got an agreement with the original figures as will be shown in -the Results. - -To check if our Python implementation produces the same results as the -implementation(s) used in the original publication, we compare each of -our figures in detail with the original figures. For this comparison, as -the original results used to generate these figures are not available to -us, we first extract the times of the spikes and unitary events from the -vector graphic image (PDF) in the original paper. Then, we directly -compare these times to the times of spikes and unitary events in our -reproduced results by plotting the former against the latter, as well as -examining the distribution of the differences between them. - -\section{Results}\label{results} - -For the reproduction of the original results in \autocite{Riehle97}, we -have to focus on reproducing Figure 2A-F and Figure 4A, since for the -rest of the figures data were either incomplete in respect to metadata -(original Figure 3) or not available (original Figure 4B,C). Figure 2 -represents the main result of the study and includes the UE analysis -that underlies the subsequent analyses. Figure 4A is an example of the -application of the UE method to data with more than 2 neurons. In terms -of complexity of the code the implementation of the UE analysis for -three or more neurons is considerably more demanding than for only two -neurons. With this example we show that our implementation is capable of -performing the UE analysis for the generic case of arbitrary number of -neurons. - -We apply our reimplementation of the UE method to preprocessed versions -of the spike train data available to us after communication with AR, -which in part were identical to those used in the original analysis. -Also, we learned from AR that Figure 2 was generated by the Matlab -implementation of the UE method while all remaining figures of the -original publication, including Figure 4A, were generated by the older -implementation in IDL (see Methods regarding versions of the original -code). - -\begin{figure}[htbp] -\centering -\includegraphics{figure1.eps} -\caption{\label{fig:figure2alignedPS}Initial attempt to reproduce Figure -2 of the original publication with trial alignment to PS. \textbf{A)} -Raster plot of two neurons (neuron 2: top of panel; neuron 3: bottom of -panel) in 32 trials (sorted identically for both neurons). \textbf{B)} -Average firing rate of each neuron calculated across trials in a sliding -window of length 100 ms in steps of 5 ms. \textbf{C)} Same raster plot -as in panel A with spike coincidences (i.e., pattern {[}1,1{]}) between -the two neurons marked by cyan squares. \textbf{D)} Empirical (cyan) and -expected (magenta) number of coincidences calculated in a time-resolved -manner (parameters of sliding window identical to panel B). \textbf{E)} -Time course of the surprise measure, calculated in same sliding windows -as in panel B. Surprise values that correspond to positive and negative -significance levels \(\alpha+=0.05\) and \(\alpha-=0.95\) are shown with -by horizontal red and green lines, respectively. \textbf{F)} Same raster -plot as in panel A with significant coincidences, i.e.~UEs, marked by -red squares.}\label{fig:figure2alignedPS} -\end{figure} - -\begin{figure}[htbp] -\centering -\includegraphics{figure2.eps} -\caption{\label{fig:figure2alignedRS}Reproduction of Figure 2 of the -original publication with trial alignment to RS. The same conventions as -in Figure ~\ref{fig:figure2alignedPS} apply to the respective -panels.}\label{fig:figure2alignedRS} -\end{figure} - -Let us start with the reproduction of Figure 2 of the original -publication. We first give a brief description of the experiment (see -the original publication for details). After the monkey was presented -with the preparatory signal (PS) he had to sit still and wait for a -response signal (RS) to start his arm movement (i.e.~equivalent to a GO -signal). The duration of the waiting period was randomly selected on a -trial-by-trial basis to be either 600, 900, 1200 or 1500 ms. In Figure 2 -of \autocite{Riehle97} only trials of the longest waiting period (1500 -ms) were used for the analysis. In these trials, times marked as -expected signals ES1, ES2, and ES3 corresponded to the ends of the three -shorter waiting periods, at which the monkey could have gotten the RS -signal but did not. As the monkey was trained to recognize and -distinguish the four waiting periods, but was not informed of the -randomly selected period for a given trial, ES1-ES3 were time points at -which the monkey expected that a signal could occur. - -Since the data file for Figure 2 contained the data as a continuous -recording of one recording session (``winny131.gdf''; 2 neurons, and -behavioral events), we extract the trials by cutting the data in a time -window around specific trigger events that belong to trials of the -longest waiting period, such that the complete trial is contained in the -cut-out. In a subsequent step, the spike times in the individual trials -are aligned to the trigger event, such that spike times in each trial -are given as relative to the trigger. - -The original publication does not provide information which event was -used as the trigger. In this experiment, 2 events that occur in every -trial could serve as trigger events, the preparatory signal PS (event -code 114) and the response signal RS (event code 124). We noticed that -the time interval between PS and RS for the longest trials was not -identical across the respective trials and varied by \(\pm\) 1 ms. Given -the UE method is applied on a time scale of 5 ms, the analysis results -therefore are expected to depend on whether trials are aligned to PS or -RS. Thus, we decide to generate the results for both alignments. - -Figures~\ref{fig:figure2alignedPS} and ~\ref{fig:figure2alignedRS} show -the results of performing the UE analysis for PS- and RS-aligned data, -respectively. Here, the analysis parameters are set to the identical -values as reported in the original publication (bin size: 5 ms, analysis -time window size: 100 ms, time step of the sliding window: 5 ms, -significance level \(\alpha= 0.05\)). The comparison of the two figures -to the original figure shows agreement in the raster displays (panel A) -and the time-resolved, trial-averaged firing rate estimates (B). -However, although the graphs of the number of coincidences per sliding -window (panel D) and the surprise measure (panel E) are similar in their -overall general behavior, they differ in the details. Thus, indeed the -choice of the alignment influences the analysis result. In order to test -if one of the two alignments is in agreement with the original -publication, we perform a detailed visual comparison of our two figures -and the original one on the basis of the spikes marked as coincident -(panels C) and as part of a UE (panel D). We notice that when aligning -to PS (Figure~\ref{fig:figure2alignedPS}), the marked spikes do not -agree in all details with the original figure. However, in -Figure~\ref{fig:figure2alignedRS}, with trials aligned to RS, we find no -disagreements with the original figure. After this visual comparison we -check if our results are exactly identical to the original ones. -Therefore, we extract the positions of the data points representing -spikes and UEs in the original figure by reading the plotting commands -in the PDF file of the original paper, and compare them to our -reproduced results. In Figure~\ref{fig:validation_fig2} A the reproduced -spike times in Figure~\ref{fig:figure2alignedRS} A are plotted against -the extracted original spike times. The plotted data points lie on the -diagonal line, indicating that the reproduced spike times correspond to -the original ones. Figure~\ref{fig:validation_fig2} B shows the same -plot for the UEs, indicating the identity of the original and reproduced -UE timings as well. To further confirm the identity, we plot the -distributions of the differences between the original and the reproduced -timings for the spikes (Figure~\ref{fig:validation_fig2} C; black) and -the UEs (Figure~\ref{fig:validation_fig2} D). The differences are at -most +/- 0.3 ms, which are considerably narrower than the +/- 1 ms -differences caused by the misaligned data shown in -Figure~\ref{fig:figure2alignedPS} A (see for comparison the gray plot in -Figure~\ref{fig:validation_fig2} C, which shows the differences between -the original spike times and the misaligned spike times). The remaining -minor differences of the spike and UE times of the correctly aligned -data and the original data are only due to slight errors in the -extraction of the spike times from the original figure, which is -inevitable because of a limited precision of the plotting commands in -the PDF file. Thus, we confirm that spike timings in the correctly -aligned data are identical to those in the original data, and our -implementation of the UE analysis applied to these data reproduces -exactly the same results as shown in the original paper. - -\begin{figure}[htbp] -\centering -\includegraphics{figure3.eps} -\caption{\label{fig:validation_fig2}\textbf{A)} The scatter plot of the -reproduced spike times in Figure~\ref{fig:figure2alignedRS} A plotted -against the extracted original spike times. \textbf{B)} Same plot as in -A but for the time of occurrences of UEs. \textbf{C)} The distributions -of the differences between the original and the reproduced timings for -spikes in Figure ~\ref{fig:figure2alignedRS} and -Figure~\ref{fig:figure2alignedPS} are shown in black and gray, -respectively. \textbf{D)} Same plot as in C but for UEs in Figure -~\ref{fig:figure2alignedRS}.}\label{fig:validationux5ffig2} -\end{figure} - -As a next step we aim at reproducing Figure 4A of \autocite{Riehle97}. -This figure contains the result of the analysis of three neurons -recorded simultaneously, in contrast to Figure 2 where only two neurons -are considered. We analyze the original data for this figure provided by -AR with the parameter values given in \autocite{Riehle97} and compare -our result to the original figure. Figure 4A of the original publication -contains the raster displays of the data in the top panel, the raster -displays with the marked coincident spikes (blue marks) in the middle -panel, and the raster displays with the marked spikes that are part of a -UE (red marks) in the bottom panel. We find that the UE result is -different, as the UEs occur at different times and between different -neurons compared to the original publication. Thus we check whether the -spike times of the individual spikes are identical between the original -and our results. Figure~\ref{fig:comparison_raster} shows a segment of -the raster plot of the original figure and the corresponding segment of -our reproduced figure. We compare the positions of the single spikes and -find that there are small discrepancies between the two raster plots in -some of the spike times. Figure~\ref{fig:comparison_raster} shows -examples of clusters of spikes marked in red that should be identical in -both raster plots but contain a few individual spikes that are slightly -shifted in our figure compared to the original figure by a very small -amount. - -\begin{figure}[htbp] -\centering -\includegraphics{PS_aligned_matlabdata_marked.eps} -\caption{\label{fig:comparison_raster}Close-ups of the original raster -display in Figure 4A of the original publication (left) and the first -data file available at hand for reproduction (right) reveal slight -differences in the positions of some spikes. Data on the right are -aligned (similar to the data on the left) to ES1 (event code 15 in the -GDF data file). The time before the alignment event is chosen as 700 ms, -and a bin size of 5 ms is used. The red marks indicate spike clusters -with identified differences between the left and the right panel, where -at least one spike is shifted in the right panel compared to the left -one.}\label{fig:comparisonux5fraster} -\end{figure} - -This leads us to the suspicion that the data are binned in a fashion -that is not consistent with the data shown in the original publication. -Personal communication with AR revealed that while Figure 2 had been -generated by the Matlab implementation of the UE analysis, Figure 4 had -been generated by the IDL implementation (see Introduction). A report by -MD written before the time of the original publication summarized a -comparison of the IDL and the Matlab implementations, and concluded that -both were correct implementations of the method, but differed in their -results due to a slightly different implementation of the down-sampling -and binning of the raw data (recorded at 10 kHz). In the workflow for -the IDL implementation, as illustrated in -Figure~\ref{fig:workflow_report} (the leftmost branch of the diagram), -the raw data were first down-sampled to a temporal resolution of 0.5 ms -(by a program \texttt{2gdf}) and then further rounded to 1 ms resolution -integer values inside the IDL implementation. The data available to us -had a resolution of 1 ms, which must have been a result of another -down-sampling procedure than the one for the IDL implementation. This -explains the difference in the raster displays, and this difference is -likely also the cause that we were initially not able to reproduce the -original UE result. - -\begin{figure}[htbp] -\centering -\includegraphics{CmpIDL_Matlab_3.eps} -\caption{\label{fig:workflow_report}Illustration of the data -preprocessing workflow, translated from the 1997 report of MD (in -German) on the comparison of the first UE implementation in IDL (left -branch) and the second implementation in Matlab -\autocite{Diesmann16_personalcomm} (right branch). Data entering both -analysis branches have a resolution of 0.1 ms (top box). In the IDL -branch spike times \(t\) in the original data are first transformed to a -resolution of \(h=0.5\) ms (by the \texttt{2gdf} program, middle left) -using the method of binning \(\lfloor t/h \rfloor\). Then the data are -read into the ``IDL UE Software'' (lower left box) and therein converted -to \(h=1\) ms resolution by the method of ``round half up'' -\(\lfloor t/h+1./2 \rfloor\) prior to analysis. Alternatively, one can -load the 0.5 ms resolution data into the ``MATLAB UE Software'' (lower -right box). Here the bin width is a parameter of the analysis and thus -data can be converted to a 1 ms resolution but results are different -from the IDL branch. Results are only identical if a prior -transformation ``T'' (diagonal in center) performs a round half up and -no further binning is done in the MATLAB program. At a later point in -time the \texttt{alexa2gdf} converter function (written in MATLAB) -became available such that data in the original 0.1 ms resolution could -directly be converted to the 1 ms resolution by binning. The full report -(``Report\_by\_MarkusDiesmann.txt'') is included in the data folder of -the repository for this paper.}\label{fig:workflowux5freport} -\end{figure} - -In our reproduction of Figure 2 of the original paper we use -preprocessed data available in 1 ms resolution, that likely experienced -the \texttt{alexa2gdf} program for conversion as shown in -Figure~\ref{fig:workflow_report} (the rightmost branch), before data are -loaded into our reimplementation of the UE analysis. However, according -to the aforementioned report by MD, we only have a chance to reproduce -Figure 4A of \autocite{Riehle97} if we have the original data or a -version of them with a time resolution lower than 1 ms available. The -original raw data with 0.1 ms resolution are presumably only available -on a storage medium and format that at present we are not able to read -and interpret. However, after we contacted AR she found the data -(``jenny201\_345\_preprocessed.gdf'') of Figure 4A of -\autocite{Riehle97} with a time resolution of 0.5 ms, which likely -experienced the \texttt{2gdf} program for conversion (middle left box in -Figure~\ref{fig:workflow_report}). - -We loaded this data at 0.5 ms resolution into Python and converted the -data from the 0.5 ms to the required 1 ms resolution by the mathematical -operation \(\left\lfloor x+\frac{1}{2}\right\rfloor\) , called -``rounding half up''. In numerical software packages, including IDL, -this operation is typically implemented by a function named round(). -However, the round() implementation of NumPy (version 1.11.0) performs -an even rounding, i.e., values exactly halfway between two integers are -rounded to the nearest even integer. Indeed, the latter implementation -of rounding did not reproduce the result of the original publication. -Thus, we used the expression \texttt{floor(x+0.5)} to perform rounding -as it is implemented by IDL. The procedure completely reproduces panel A -of Figure 4 in the original publication (see -Figure~\ref{fig:reproducedFig4A}). - -\begin{figure}[htbp] -\centering -\includegraphics{figure6.eps} -\caption{\label{fig:reproducedFig4A}Reproduction of Figure 4A of the -original publication. The left part of the figure shows the UE analysis -result for the data aligned to ES1 (event code 15 in the GDF data file) -with a time before the event (pre-time) set to 699 ms. The right part of -the figure shows the analysis of the same data aligned to RS and with a -pre-time of 99 ms. The left and right parts of the figure show 96 and -128 trials, respectively.}\label{fig:reproducedFig4A} -\end{figure} - -\section{Conclusion}\label{conclusion} - -We are able to reproduce the original results of \autocite{Riehle97} by -applying a new reimplementation of the Unitary Events analysis method in -Python to the original data. The method involves a number of numerical -computations and is very sensitive, as we show here by the comparison of -Figures ~\ref{fig:figure2alignedPS} and ~\ref{fig:figure2alignedRS}, -which differed in the events that the trials were aligned to. This -difference in the alignment would not affect the results if the time -difference between the two events (PS and RS) were identical across the -trials. But since the latter was not the case due to hardware features -of the recording setup (as we learned from the first author of the -original publication), the binning of the data started at a slightly -different time points in different trials. This likely led to a loss or -an addition of a spike in a bin and thus to a small difference of the -number of spike synchrony events (see also the discussion on the issues -of exclusive binning in \autocite{Gruen99}). In spite of this -sensitivity of the method, we succeeded in generating results that are -exactly identical to the original ones, as confirmed by the direct -comparison of the positions of spikes and UEs in the original and the -reproduced figures. This is a strong indication that our new -implementation of the analysis faithfully implemented the UE method. - -The event to which the data were aligned and the cut time which then -defined the start of the (exclusive) binning was not documented in the -original publication. Also the original scripts for the analysis are not -available anymore, which could have revealed this information, even -without having the original UE software code at hand. Thus, due to the -lack of documentation we are only able to reproduce the results of -\autocite{Riehle97} by communicating with some of the authors of the -original publication. - -The reproduction of Figure 4A of the original publication is a further -and important test whether our reimplementation is also correct for -N\textgreater{}2 neurons. This is relevant since the implementation -requires a more generic, complex algorithm for the analysis than the one -that can be used for only N=2. In the case of two neurons, there is only -one pattern type which has to be analyzed (i.e. {[}1,1{]}). However, in -the case of e.g.~3 neurons there are already 4 different spike patterns -to analyze ({[}1,1,0{]}, {[}0,1,1{]}, {[}1,0,1{]}, and {[}1,1,1{]}), and -even much more for more neurons (\(2^{N}-N-1\)). The statistics of each -of the patterns is performed separately and, therefore, the bookkeeping -needs to be carefully done. - -The reproduction of Figure 4A is more complicated than reproducing -Figure 2 due to additional reasons. First of all, the data were not -available to us. After requesting them from the first author of the -original publication we received data and were not able to reproduce the -result - in terms of the UE results, but also the data seemed slightly -different. After further consultation with the original authors we -learned that the original Figure 4A was not generated by the Matlab -implementation used for Figure 2 but by another implementation in IDL. -Both are not available to us. However, we were told that the third -author, MD, of the original publication performed a thorough comparison -of the two implementations at the time and the final report on that -investigation was made available to us. This enables us to define the -correct workflow that reproduces the original result, given we have the -data in the correct resolution at hand. - -The reimplemented UE analysis software contains extensions for improving -the statistics that were developed after the original publication. On -the one hand, it contains the option to adjust the statistics to take -into account cross-trial inhomogeneity by calculating the number of -expected spike synchrony events based on the firing rates in a -trial-by-trial fashion (option: \emph{``analytic\_TrialByTrial''}) as -suggested in \autocite{Gruen03b}, in contrast to using trial averages of -the firing rates. On the other hand, our reimplementation offers the -possibility to calculate the significance of the empirical number of -spike synchrony events based on a Monte Carlo approach (option: -\emph{``surrogate\_TrialByTrial''}). Instead of computing the -significance using a parametric distribution based on the estimate of -the firing rates, the null-hypothesis of independence is implemented by -surrogate data \autocites{Gruen09}{GruenRotter10_Chap10}{Louis10}. By -repeated intentional manipulation of the original data, potential spike -synchrony is deleted. Each of these surrogate data created by this -procedure are then searched - as the original data - for spike -synchrony, and these numbers create the distribution underlying the -significance test of the method. Obviously this version is considerably -more computationally expensive than the parametric approach used here -for the reproduction of \autocite{Riehle97} and we are currently working -on an HPC implementation to make use of parallelization. The Python -implementation of the UE method is publicly available in the open source -software package Elephant at http://neuralensemble.org/elephant/. - -If the authors of the original paper would not have been accessible, we -would not have been able to reproduce the results. Nevertheless, here -our final validation of the reproduction is based on the values -extracted from the vector graphic image (PDF) in the original paper. In -an optimal scenario, we would be able to exactly validate the results -based on a numerical comparison. To do so, all of the following pieces -of information would have had to be available at hand: - -\begin{enumerate} -\def\labelenumi{\arabic{enumi}.} -\item - the original primary data -\item - metadata describing the primary data in detail -\item - the original statistics software package (e.g.~Unitary Events) -\item - the loading routine for the data -\item - all specific code required to produce each figure of the original - publication -\item - detailed documentation of all code -\item - the original software environment (with programs available in the - original versions used), including, e.g.~the interpreter/compiler - (here: Matlab) and operating system -\item - unique identifiers of the data records that unambiguously identify - data from within the analysis code -\end{enumerate} - -In the analysis presented in this work, not even the original primary -data (1), recorded more than 20 years ago, but only a slightly -preprocessed version is available. However, even today many of the -pieces of information listed above are often not made available by -scientists. In part, this is due to the enormous complexity of the task -to record all information in fine detail leading from the experiment to -an analysis result. Moreover, there is still a lack of software tools to -support researchers in the process of acquiring, storing, and organizing -this information. Currently, there are emerging approaches suggested for -metadata annotation (2) of electrophysiological data, such as the -odML\footnote{\url{https://github.com/G-Node/python-odml}} framework -(see, e.g. \autocites{Grewe2011}{Zehl2016}) for storing hierarchical -collections of metadata or the NIX\footnote{\url{https://github.com/G-Node/nix/wiki}} -data format \autocite{Adrian2014} for linking data and metadata. In our -concrete example, the information about the hardware limitations in -storing the event times, would have been essential information contained -in the metadata. Using modern tools for version control, points (3)-(6) -can be easily addressed. There are emerging approaches to keep software -environment, i.e., the original Matlab version and the operating system -(7), e.g., by freezing the environment in a virtual machine. Point (8) -is still challenging, because it requires the data to be addressed in an -unambiguous manner from within the analysis scripts. Including data -within the code repositories is typically prohibitive due to the size of -the data. A solution would be to deposit data in public or private -databases that allow data to be identified using a unique identifier in -combination with a tool to generate a detailed provenance track of the -analysis process, but the implementation of tools and services for the -workflows used in data analysis of electrophysiological data is still an -ongoing endeavor \autocites{badia_incf_2015}{Denker2015_000}. In -summary, there are still components missing such that researchers are -put into a position to build complex data acquisition and analysis -workflows that enable optimal reproducibility in neuroscience. - -\section{Author Contribution}\label{author-contribution} - -\textbf{VR}: Implementing the UE analysis in Python, Reproduction of the -figures, Discussion of the results, Drafting of the manuscript, Revision -of the draft. \textbf{JI}: Reproduction of the figures (with VR), -Discussion of the results (with all other authors), Drafting of the -manuscript (with VR), Revision of the draft (with all other authors). -\textbf{MDe}: Drafting of the manuscript (with VR), Revision of the -draft (with all other authors), Assisted in integrating the UE analysis -in the Elephant library. \textbf{SG}: Conceived the original idea of -reproduction, mediated the communications with original authors, -provided suggestions for correcting errors in intermediate reproduction -results, discussed the results, revised the manuscript (with all other -authors). - -\section{Acknowledgements}\label{acknowledgements} - -This project received funding from EU Grant 720270 (HBP), Deutsche -Forschungsgemeinschaft Grant DE 2175/2-1 and GR 1753/4-2 of the Priority -Program (SPP 1665), the German-Japanese Computational Neuroscience -Project (German Federal Ministry for Education and Research, BMBF Grant -01GQ1114), from the Helmholtz Portfolio Theme ``Supercomputing and -Modeling for the Human Brain'', and from the Osaka Univ for the project -`Neural mechanism of active vision studied by combining large-scale -sampling of neural activity and advanced computational analysis'. - -We thank Alexa Riehle and Markus Diesmann for fruitful discussions. - -{\sffamily \small - \printbibliography[title=References] -} -\end{document} diff --git a/article/rescience-template.tex b/article/rescience-template.tex index 01b14d5..b9b2abf 100644 --- a/article/rescience-template.tex +++ b/article/rescience-template.tex @@ -121,7 +121,7 @@ \sffamily \ReScience~$$\vert$$ \href{http://rescience.github.io}{rescience.github.io} \hypersetup{urlcolor=blue}} -\fancyfoot[C]{\sffamily \thepage} +\fancyfoot[C]{\sffamily $Publication.number$ - \thepage} \fancyfoot[R]{\sffamily $Publication.date$ $$\vert$$ Volume $Publication.volume$ $$\vert$$ Issue $Publication.issue$}