Skip to content

Latest commit

 

History

History
133 lines (118 loc) · 11.4 KB

README.md

File metadata and controls

133 lines (118 loc) · 11.4 KB
Raw

bbi-genome-data

Overview

bbi-genome-data consists of bash scripts and programs for preparing genome-related files for the bbi-scirna* and bbi-sciatac-* processing pipelines.

Prerequisites

  1. As the genome building pipeline is run interactively, please use a terminal multiplexer such as tmux. tmux sessions are persistent which means that programs in tmux will continue to run even if you get disconnected. You can start a tmux session by using:
module load tmux/latest
tmux

If you get disconnected, you can return to the head node you were using (grid-head1 or grid-head2) and type:

tmux attach -t session-name

which will return you to your session. See a handy tmux tutorial here.

  1. Always start with a qlogin session before you begin the pipeline. This can be done by
qlogin -l mfree=20G

Build programs

Once you have a qlogin session running, clone the repository bbi-genome-data and build the required programs using:

git clone https://github.com/bbi-lab/bbi-genome-data.git
cd bbi-genome-data/src/sequtil
make fasta_getseqs
make libsquid md5 md5_seq

Setup

  • move the programs fasta_getseqs and md5_seq, which you made in src/seq_util, to a convenient directory, if desired
  • move the files R/generate_gene_body_file.R and R/generate_tss_file.R to a convenient directory, if desired
  • move the bash scripts in the scripts directory to a convenient directory, if desired
  • edit the script genome.02.definitions.sh, as required
    • set the variables MD5_SEQ and FASTA_GETSEQS to the paths for the fasta_getseqs and md5_seq programs
    • set the variables SAMTOOLS and BEDTOOLS to the paths for the samtools and bedtools programs
  • edit the script atac.02.definitions.sh as required
    • set the variable BOWTIE2_BUILD to the path for the bowtie2-build program
    • set the variables R_GENERATE_TSS_FILE and R_GENERATE_GENE_BODY_FILE to the paths for the generate_tss_file.R and generate_gene_body_file.R scripts
    • set the variable BEDTOOLS to the path for the bedtools program
  • edit the script all.02.definitions.sh to set the variable STAGE_DIR, where the scripts will make their output directories.
  • install James Kent's faCount utility in a convenient directory and edit scripts/atac.02.definitions.sh to set the variable FACOUNT to its path. Source code and binaries for faCount are available from http://hgdownload.soe.ucsc.edu/admin/exe/ and https://github.com/ucscGenomeBrowser/kent/tree/master/src

Make organism-specific definition files.

  • organism-specific definition files are stored in the bbi-genome-data/organisms directory
  • copy the file organisms/genome.template.sh to genome.<organism>.sh and edit it as required for the organism. The variables are described in genome.template.sh.

Run the scripts

  • run genome.01.run.sh <organism> to download and prepare the genome fasta and gtf files for <organism>
  • run genome.01.run_barnyard.sh to prepare files for the human-mouse barnyard processing
  • run rna.01.run.sh <organism> to prepare files for the sci-RNA-seq processing pipeline.
  • run rna.01.run.sh barnyard to prepare human-mouse barnyard files for the sci-RNA-seq processing pipeline.
  • run atac.01.run.sh <organism> to prepare files for the sci-ATAC-seq processing pipeline for <organism>
  • run atac.01.run.sh barnyard to prepare human-mouse barnyard files for the sci-ATAC-seq processing pipeline

Note: All of the scripts above are meant to be run from a qlogin session only. Submitting any of these as a job to the cluster (via qsub) will generate an error. For example, run the script genome.01.run.sh <organism> as follows:

./genome.01.run.sh <organism>

Check log files

  • review the files <output_dir>/log.out. Key information is marked with the string CHECKPOINT.
  • read the script functions to understand the reports in <output_dir>/log.out
  • the files <output_dir>/record.out have particularly important information for record-keeping (which is in log.out too).

Notes:

  • script names have the form <pipeline>.<number>.<description>.sh
    • scripts <pipeline> values
      • all scripts are sourced in scripts for genome, rna, and atac
      • genome scripts download and process fasta and gtf files used in both the sci-RNA-seq and sci-ATAC-seq processing pipelines
      • rna scripts make files related to the sci-RNA-seq processing pipeline
      • atac scripts make files related to the sci-ATAC-seq processing pipeline
    • script <number> values
      • 01 scripts are executable
      • 02 scripts have variable and function definitions for either all pipelines or a specific pipeline
      • 03+ scripts have functions used in a pipeline. The numbers give the order in which the functions are used in the 01 scripts
      • 10 scripts have functions for cleaning up by deleting unnecessary files
  • consider setting STAGE_DIR to a temporary staging directory where the genome directories and their files are made and checked. Then move them to the directory where they will be used.
  • the scripts make a directory in which they write their output files
    • the scripts make their directory in the parent directory specified in the variable STAGE_DIR, which is defined in the script all.02.definitions.sh
    • the script genome.01.run.sh makes the directory <organism>_gsrc
    • the script genome.01.run_barnyard.sh makes the directory barnyard_gsrc
    • the script rna.01.run.sh makes the directories <organism> and <organism>_star
    • the script atac.01.run.sh makes the directory <organism>_atac
  • <output_dir> refers to <organism>_gsrc, barnyard_gsrc, and <organism>_atac
  • the scripts write diagnostic information to the terminal and to the file <output_dir>/log.out
  • the script genome.01.run.sh writes the fasta files
    • *.fa.gz is the compressed fasta file downloaded from Ensembl
    • *.fa is the uncompressed Ensembl fasta file
    • *.fa.filtered has unmodified sequences selected from the Ensembl fasta file
    • *.fa.finished is the *.fa.filtered file with pseudo-autosomal regions (PARs) masked and/or manual modifications, if either is applied; otherwise, it is the same as *.fa.filtered
    • *.fa.finished.bz2 is the *.fa.finished file with bzip2 compression
  • if necessary, make manual edits to the *.fa.finished file and compress it with bzip2 -k. For example, in the Ensembl release 99, the Macaca mulatta genome does not include the mitochondrial chromosome so you may want to add it to the *.fa.finished fasta file. You may want to describe manual modifications in a 'TAG' prefixed line added to the log.out file
  • custom genome preparation
    • download the base FASTA and GTF files to a source directory
    • add, remove, or modify the sequences in the FASTA file and edit the GTF file accordingly. The header for sequences that you want included in the finished FASTA file must have the token 'REF' in the fourth field, when you select sequences using the sequences_to_keep_ref() function.
    • make an organism file for the custom genome in which
      • ENSEMBL_DNA_URL points to the source directory with the modified FASTA file
      • ENSEMBL_GTF_URL points to the source directory with the modified GTF file
      • FASTA_GZ is the name of the modified, compressed FASTA file
      • GTF_GZ is the name of the modified, compressed GTF file
      • WGET_FASTA_GZ=0
      • WGET_GTF_GZ=0
    • continue by running the genome.01.run.sh, rna.01.run.sh, and atac.01.run.sh scripts
  • Ensembl fasta files
  • there is information on selecting and processing genome sequences at http://lh3.github.io/2017/11/13/which-human-reference-genome-to-use/ and https://www.biostars.org/p/342482/
  • how to select fasta sequences from the Ensembl fasta file
    • the functions sequences_to_keep_ref() and sequences_to_keep_named() make a file called sequences_to_keep.txt that has the names of the sequences to copy from the Ensembl fasta to *.fa.filtered
    • sequences_to_keep_ref() chooses sequences with the word REF as the 4th token in the header and writes the sequence names to the file sequences_to_keep.txt
    • sequences_to_keep_named() copies the sequence names listed in the variable SEQUENCES_TO_KEEP_ALIGNER, which is defined in the genome.<organism>.sh file, to the file sequences_to_keep.txt
    • use either sequences_to_keep_ref() or sequences_to_keep_named() in genome.01.run.sh but not both
  • the barnyard genome files are built from the human and mouse files in the human_gsrc and mouse_gsrc directories so run genome.01.run.sh for human and mouse (and preserve the fasta files in them) and then run genome.01.run_barnyard.sh
  • there are functions for cleaning unwanted files from the <output_dir> directories but they do nothing as distributed. Edit them in order to delete unwanted files
  • pseudo-autosomal regions: finding PAR coordinates for organisms seems difficult. Perhaps, they are defined for only a few organisms. I found human and mouse PAR coordinates at https://ftp.ncbi.nlm.nih.gov/genomes/all/GCF/000/001/405/GCF_000001405.39_GRCh38.p13/GCF_000001405.39_GRCh38.p13_assembly_regions.txt and https://ftp.ncbi.nlm.nih.gov/genomes/all/GCF/000/001/635/GCF_000001635.26_GRCm38.p6/GCF_000001635.26_GRCm38.p6_assembly_regions.txt, respectively. The PARs appear to be masked in the Y chromosome
  • ATAC-specific issues
  • I feel ambivalent about these scripts. I want to automate downloading, preparing, and checking the files. However, checking the resulting output files requires some care and I am concerned that the automation, coupled with the large amount of diagnostic information in the log.out files, may overwhelm the user and result in some complacency. Additionally, I strive to eliminate code duplication in the scripts, which results in unfortunate complexity. I am not confident that I found an effective balance in these scripts.