From 609b921035cc7e886cd2b0ee7020d9a40f95e7c2 Mon Sep 17 00:00:00 2001
From: GuillaumeG <guipagui@gmail.com>
Date: Tue, 17 Sep 2024 18:02:30 +0200
Subject: [PATCH] `doe_synbio` update version (#195)

* feat(tools): doe_synbio, update newer version

* fix(tools): icfree, package name and test 2 of sampler

* chore(gihub): add style for R lint and update skipping files

* fix(tools): doe_synbio, add default value to not block shed_update
---
 .github/styler.R                              | 36 +++++++++
 .gitignore                                    |  1 +
 tools/doe_synbio/macros.xml                   |  1 -
 tools/doe_synbio/optdoe.xml                   |  2 +-
 tools/doe_synbio/sampler.xml                  | 78 +++++++++++++++++--
 .../sampler/output/sampling.test-3.tsv        |  4 +
 .../sampler/output/sampling.test-4.tsv        |  5 ++
 7 files changed, 117 insertions(+), 10 deletions(-)
 create mode 100755 .github/styler.R
 create mode 100644 tools/doe_synbio/test-data/sampler/output/sampling.test-3.tsv
 create mode 100644 tools/doe_synbio/test-data/sampler/output/sampling.test-4.tsv

diff --git a/.github/styler.R b/.github/styler.R
new file mode 100755
index 00000000..c5c0cb5d
--- /dev/null
+++ b/.github/styler.R
@@ -0,0 +1,36 @@
+#!/usr/bin/env Rscript
+
+library("argparse")
+library("styler")
+
+parser <- ArgumentParser(description = "Call styler")
+parser$add_argument("dir",
+    metavar = "DIR", type = "character",
+    help = "File to parse"
+)
+parser$add_argument("--dry",
+    choices = c("off", "on"), default = "on"
+)
+args <- parser$parse_args()
+
+file_info <- file.info(args$dir)
+is_directory <- file_info$isdir
+
+if (is_directory) {
+    captured_output <- capture.output({
+        result <- style_dir(args$dir, indent_by = 4, dry = args$dry, recursive = TRUE)
+    })
+} else {
+    captured_output <- capture.output({
+        result <- style_file(args$dir, indent_by = 4, dry = args$dry)
+    })
+}
+
+n <- nrow(subset(result, changed == TRUE))
+if (n > 0) {
+    if (args$dry == "off") {
+        print(paste("Changed", n, "files"))
+    } else {
+        stop(paste("Linting failed for", n, "files"))
+    }
+}
diff --git a/.gitignore b/.gitignore
index 9bfb4d39..da400fd2 100644
--- a/.gitignore
+++ b/.gitignore
@@ -6,3 +6,4 @@ node_modules
 tool_test_output.html
 tool_test_output.json
 .hg_archival.txt
+*.swp
diff --git a/tools/doe_synbio/macros.xml b/tools/doe_synbio/macros.xml
index 11595e04..c89095dd 100644
--- a/tools/doe_synbio/macros.xml
+++ b/tools/doe_synbio/macros.xml
@@ -1,6 +1,5 @@
 <macros>
     <token name="@VERSION_SUFFIX@">0</token>
-    <token name="@LICENSE@">MIT</token>
     <xml name="stdio">
         <stdio>
             <regex match="WARNING" level="warning" />
diff --git a/tools/doe_synbio/optdoe.xml b/tools/doe_synbio/optdoe.xml
index 53eb750c..2fe63e7e 100644
--- a/tools/doe_synbio/optdoe.xml
+++ b/tools/doe_synbio/optdoe.xml
@@ -1,4 +1,4 @@
-<tool id="doe_synbio_optdoe" name="Optimal Design of Experiment" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" license="@LICENCE@">
+<tool id="doe_synbio_optdoe" name="Optimal Design of Experiment" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" license="MIT">
     <description>Combine selected genetic parts and enzyme variants for the desired pathways</description>
     <macros>
         <import>macros.xml</import>
diff --git a/tools/doe_synbio/sampler.xml b/tools/doe_synbio/sampler.xml
index e8ffd5e5..50662dd1 100644
--- a/tools/doe_synbio/sampler.xml
+++ b/tools/doe_synbio/sampler.xml
@@ -1,28 +1,63 @@
-<tool id="doe_synbio_sampler" name="iCFree sampler" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" license="@LICENCE@">
+<tool id="doe_synbio_sampler" name="iCFree sampler" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" license="MIT">
     <description>Generate Latin Hypercube Samples for given components.</description>
     <macros>
         <import>macros.xml</import>
-        <token name="@TOOL_VERSION@">2.4.0</token>
+        <token name="@TOOL_VERSION@">2.5.0</token>
     </macros>
     <requirements>
-        <requirement type="package" version="@TOOL_VERSION@">icfree-ml</requirement>
+        <requirement type="package" version="@TOOL_VERSION@">icfree</requirement>
     </requirements>
     <expand macro="stdio"/>
     <command detect_errors="exit_code"><![CDATA[
+        #import json
+        #set $components = {}
+        #for $x in $adv.rep
+            $components.__setitem__(str($x.component_name), str($x.component_value))
+        #end for
+
         python -m icfree.sampler
             '$input_file'
             '$output_sampling'
             '$num_samples'
-            --step '$adv.step'
+            #if str($adv.range_cond.range_param) == 'ratios'
+                --ratios '$adv.range_cond.by_ratios'
+            #else
+                --step '$adv.range_cond.by_step'
+            #end if
             #if str($adv.seed_cond.seed_param) == 'not_random'
                 --seed '$adv.seed_cond.seed'
             #end if
+            #if len(components) > 0
+                --fixed_values '#echo json.dumps(components)#'
+            #end if
     ]]></command>
     <inputs>
         <param name="input_file" type="data" format="tabular" label="Input file with components and their max values." />
-        <param name="num_samples" type="integer" value="99" min="1" max="198" label="Number of samples to generate" />
+        <param name="num_samples" type="integer" value="1" min="1" label="Number of samples to generate" />
         <section name="adv" title="Advanced Options" expanded="false">
-            <param name="step" type="float" value="2.5" min="1" max="10" label="Step size for creating discrete ranges" />
+            <!-- Range -->
+            <conditional name="range_cond">
+                <param name="range_param" type="select" label="Discrete range" help="How to create the discrete range">
+                    <option value="step" selected="true">step</option>
+                    <option value="ratios">ratios</option>
+                </param>
+                <when value="step">
+                    <param name="by_step" type="float" value="2.5" min="0" label="Step size for creating discrete ranges" />
+                </when>
+                <when value="ratios">
+                    <param name="by_ratios" type="text" label="Ratios to be sampled">
+                        <validator type="regex" message="List of float values, separated by a comma without of spaces">^(\d+(\.\d+)?)(,\d+(\.\d+)?)*$</validator>
+                    </param>
+                </when>
+            </conditional>
+            <!-- Fixed values -->
+            <repeat name="rep" title="Fixed values for components as a dictionary">
+                <param name="component_name" type="text" value="" label="Component Name">
+                    <validator type="empty_field" message="Not empty" />
+                </param>
+                <param name="component_value" type="integer" value="1" min="1" label="Value" />
+            </repeat>
+            <!-- Seed -->
             <conditional name="seed_cond">
                 <param name="seed_param" type="select" label="Seed" help="Choose a seed or let it as random">
                     <option value="random" selected="true">random</option>
@@ -50,15 +85,42 @@
             <param name="seed" value="0" />
             <output name="output_sampling" file="sampler/output/sampling.test-1.tsv" ftype="tabular" compare="diff" />
         </test>
-        <!-- test 1: check if identical outputs are produced with default parameters  -->
+        <!-- test 2: with specific step  -->
         <test>
             <param name="input_file" value="sampler/input/components.tsv" />
             <param name="num_samples" value="4" />
-            <param name="step" value="3.1" />
+            <param name="range_param" value="step" />
+            <param name="by_step" value="3.1" />
             <param name="seed_param" value="not_random" />
             <param name="seed" value="0" />
             <output name="output_sampling" file="sampler/output/sampling.test-2.tsv" ftype="tabular" compare="diff" />
         </test>
+        <!-- test 3: with fixed values  -->
+        <test>
+            <param name="input_file" value="sampler/input/components.tsv" />
+            <param name="num_samples" value="3" />
+            <param name="seed_param" value="not_random" />
+            <param name="seed" value="0" />
+            <repeat name="rep">
+                <param name="component_name" value="Hela lysate" />
+                <param name="component_value" value="10" />
+            </repeat>
+            <repeat name="rep">
+                <param name="component_name" value="RNA 1ug/uL" />
+                <param name="component_value" value="20" />
+            </repeat>
+            <output name="output_sampling" file="sampler/output/sampling.test-3.tsv" ftype="tabular" compare="diff" />
+        </test>
+        <!-- test 4: with ratios  -->
+        <test>
+            <param name="input_file" value="sampler/input/components.tsv" />
+            <param name="num_samples" value="4" />
+            <param name="seed_param" value="not_random" />
+            <param name="seed" value="0" />
+            <param name="range_param" value="ratios" />
+            <param name="by_ratios" value="0.5,0.1,0.3,0.2" />
+            <output name="output_sampling" file="sampler/output/sampling.test-4.tsv" ftype="tabular" compare="diff" />
+        </test>
     </tests>
     <help><![CDATA[
 Sampler
diff --git a/tools/doe_synbio/test-data/sampler/output/sampling.test-3.tsv b/tools/doe_synbio/test-data/sampler/output/sampling.test-3.tsv
new file mode 100644
index 00000000..e4f8d27a
--- /dev/null
+++ b/tools/doe_synbio/test-data/sampler/output/sampling.test-3.tsv
@@ -0,0 +1,4 @@
+Hela lysate,Access prot 50%,Reaction mix,RNA 1ug/uL
+10.0,165.0,257.5,20.0
+10.0,325.0,192.5,20.0
+10.0,310.0,305.0,20.0
diff --git a/tools/doe_synbio/test-data/sampler/output/sampling.test-4.tsv b/tools/doe_synbio/test-data/sampler/output/sampling.test-4.tsv
new file mode 100644
index 00000000..574429ea
--- /dev/null
+++ b/tools/doe_synbio/test-data/sampler/output/sampling.test-4.tsv
@@ -0,0 +1,5 @@
+Hela lysate,Access prot 50%,Reaction mix,RNA 1ug/uL
+200.0,80.0,120.0,20.0
+200.0,80.0,40.0,100.0
+500.0,120.0,40.0,60.0
+300.0,80.0,120.0,20.0