Merge pull request #521 from hechth/hechth/issue427

Isolib tool first version

tools/isolib/.shed.yml

@@ -0,0 +1,11 @@
+categories:
+- Metabolomics
+description: Create an isotopic pattern library for given compounds and adducts.
+homepage_url: https://github.com/RECETOX/galaxytools/
+long_description: |
+  Create MSP files containing the isotopic patterns for given molecules with given adducts.
+  The tool is based on enviPat and the RforMassSpectrometry toolbox.
+name: isolib
+owner: recetox
+remote_repository_url: https://github.com/RECETOX/galaxytools/tree/master/tools/isolib
+type: unrestricted

tools/isolib/isolib.R

@@ -0,0 +1,80 @@
+library(enviPat)
+library(Spectra)
+library(MsBackendMsp)
+library(MetaboCoreUtils)
+
+#' @param args A list of command line arguments.
+main <- function() {
+  data(isotopes)
+  data(adducts)
+
+  args <- commandArgs(trailingOnly = TRUE)
+  compound_table <- read.delim(args[1], stringsAsFactors = FALSE)
+  adducts_to_use <- c(unlist(strsplit(args[2], ",", fixed = TRUE)))
+
+  chemforms <- compound_table$formula
+  chemforms <- check_chemform(isotopes, chemforms)[, 2]
+
+  spectra <- data.frame()
+
+  for (current in adducts_to_use) {
+    adduct <- adducts[adducts$Name == current, ]
+    multiplied_chemforms <- multiform(chemforms, adduct$Mult)
+
+    if (adduct$Ion_mode == "negative") {
+      merged_chemforms <- subform(multiplied_chemforms, adduct$Formula_ded)
+    } else {
+      merged_chemforms <- mergeform(multiplied_chemforms, adduct$Formula_add)
+    }
+
+    charge_string <- paste0(if (adduct$Charge > 0) "+" else "-", if (abs(adduct$Charge) > 1) abs(adduct$Charge) else "")
+    adduct_string <- paste0("[", adduct$Name, "]", charge_string)
+    precursor_mz <- calculateMass(multiplied_chemforms) + adduct$Mass
+
+    if (args[4] == TRUE) {
+      names <- paste(compound_table$name, paste0("(", adduct$Name, ")"), sep = " ")
+    } else {
+      names <- compound_table$name
+    }
+
+    spectra_df <- data.frame(
+      name = names,
+      adduct = adduct_string,
+      formula = chemforms,
+      charge = adduct$Charge,
+      ionization_mode = adduct$Ion_mode,
+      precursor_mz = precursor_mz,
+      msLevel = as.integer(1)
+    )
+
+    if ("rt" %in% colnames(compound_table)) {
+      spectra_df$retention_time <- compound_table$rt
+    }
+
+    patterns <- enviPat::isopattern(
+      isotopes = isotopes,
+      chemforms = merged_chemforms,
+      charge = adduct$Charge,
+      threshold = as.numeric(args[3]),
+    )
+
+    mzs <- list()
+    intensities <- list()
+    for (i in seq_along(patterns)) {
+      mzs <- append(mzs, list(patterns[[i]][, 1]))
+      intensities <- append(intensities, list(patterns[[i]][, 2]))
+    }
+
+    spectra_df$mz <- mzs
+    spectra_df$intensity <- intensities
+    spectra <- rbind(spectra, spectra_df)
+  }
+
+  sps <- Spectra(spectra)
+  export(sps, MsBackendMsp(), file = args[5])
+}
+
+# Get the command line arguments
+args <- commandArgs(trailingOnly = TRUE)
+# Call the main function
+main()

tools/isolib/isolib.xml

@@ -0,0 +1,72 @@
+<tool id="isolib" name="isolib" version="1.0.0+galaxy0" profile="21.09">
+    <description>create an isotopic pattern library for given compounds and adducts</description>
+    <creator>
+        <person
+            givenName="Helge"
+            familyName="Hecht"
+            url="https://github.com/hechth"
+            identifier="0000-0001-6744-996X" />
+        <organization
+            url="https://www.recetox.muni.cz/"
+            email="GalaxyToolsDevelopmentandDeployment@space.muni.cz"
+            name="RECETOX MUNI" />
+    </creator>
+    <edam_operations>
+        <edam_operation>operation_3632</edam_operation>
+    </edam_operations>
+    <requirements>
+        <requirement type="package" version="1.10.0">bioconductor-metabocoreutils</requirement>
+        <requirement type="package" version="1.12.0">bioconductor-spectra</requirement>
+        <requirement type="package" version="1.6.0">bioconductor-msbackendmsp</requirement>
+        <requirement type="package" version="2.6">r-envipat</requirement>
+    </requirements>
+    <command detect_errors="exit_code"><![CDATA[
+        Rscript ${__tool_directory__}/isolib.R '${input_file}' '${adducts}' '${threshold}' '${append_adduct}' '${isotope_library}'
+    ]]></command>
+    <inputs>
+        <param name="input_file" type="data" format="tabular" label="Table with input compounds"/>
+        <conditional name="ionization">
+            <param name="ionization" type="select" label="Ionization mode" help="Ionization mode used in the experiment">
+                <option value="negative" selected="true">negative</option>
+                <option value="positive">positive</option>
+            </param>
+            <when value="positive">
+                <param name="adducts" type="select" label="Adducts" multiple="true" help="Adducts to use">
+                    <option value="M+H" selected="true">M+H</option>
+                </param>
+            </when>
+            <when value="negative">
+                <param name="adducts" type="select" label="Adducts" multiple="true" help="Adducts to use">
+                    <option value="M-H" selected="true">M-H</option>
+                    <option value="M-2H">M-2H</option>
+                    <option value="2M-H">2M-H</option>
+                </param>
+            </when>
+        </conditional>
+        <param name="threshold" type="float" min="0" max="100" value="1" label="Threshold" help="Probability threshold to use as cutoff for isotopic pattern distribution - this can be used to remove low abundant peaks and improve computation performance." />
+        <param name="append_adduct" type="boolean" truevalue="TRUE" falsevalue="FALSE" checked="true" label="Append adduct to compound name" help="Append the adduct string to the compound name for easy identification." />
+    </inputs>
+    <outputs>
+        <data format="msp" name="isotope_library"/>
+    </outputs>
+
+    <tests>
+        <test>
+            <param name="input_file" value="lc_markers_neg.tsv"/>
+            <output name="isotope_library" file="test0.msp"/>
+        </test>
+    </tests>
+    <help><![CDATA[
+        This tool computes isotopic patterns for given compounds and adduct forms.
+        The compound table input file should contain the following columns:
+        - name: compound name
+        - formula: compound formula
+        - rt (optional): retention time
+
+        The output is a spectral library in the MSP format.
+    ]]></help>
+    <citations>
+        <citation type="doi">10.1021/acs.analchem.5b00941</citation>
+        <citation type="doi">10.3390/metabo12020173</citation>
+    </citations>
+</tool>

tools/isolib/test-data/lc_markers_neg.tsv

@@ -0,0 +1,24 @@
+formula	name	rt
+C8H6Cl2O3	2,4-Dichlorophenoxyacetic acid ou 2,4-D	484.2
+C9H15N3O1	2-diethylamino-6-methyl pyrimidin-4-ol/one	451.8
+C5H2Cl3N1O1	3,5,6-Trichloro-2-pyridinol	499.2
+C13H10O3	3-phenoxybenzoic acid	517.8
+C13H9FO3	4-Fluoro-3-phenoxybenzoic acid	532.2
+C6H5NO3	4-nitrophenol	165
+C6H4Cl1N1O2	6-Chloronicotinic acid	172.2
+C19H28N2O5S	Acetochlor mercapturate	607.2
+C19H28N2O5S1	Alachlor mercapturate	607.2
+C10H12N2O3S1	Bentazone	747.6
+C4H11O3P1S1	Diethylthiophosphate	177
+C14H17Cl2NO2	Fenhexamid	689.4
+C11H13ClO2	Fenvalerate free acid	585
+C12H4Cl2F6N4OS	Fipronil	750
+C12H4Cl2F6N4O2S	Fipronil sulfone	771
+C16H22ClN3O2	Hydroxy-tebuconazole	696.6
+C16H11ClF6N2O	Fluopyram	725.4
+C10H11Cl1O3	Mecoprop	544.8
+C7H9NO2S	p-Toluenesulfonamide	362.4
+C12H7Cl3O2	Triclosan	811.8
+C18H15Cl3O8	Triclosan glucuronide	665.4
+C12H7Cl3O5S	Triclosan sulfate	695.4
+C9H9N4Cl	acetamiprid-N-desmethyl	402.6