Mercurial > repos > theo.collard > ballgown_wrapper
view ballgown.xml @ 1:fa62657e9b57 draft
Uploaded
author | theo.collard |
---|---|
date | Wed, 26 Apr 2017 08:29:56 -0400 |
parents | |
children | 05977e96375b |
line wrap: on
line source
<tool id="ballgown" name="Ballgown" version="0.5.0" workflow_compatible="true"> <description>Flexible, isoform-level differential expression analysis</description> <requirements> <requirement type="package" version="2.2.0">bioconductor-ballgown</requirement> <requirement type="package" version="0.5.0">r-dplyr</requirement> <requirement type="package" version="1.3.2">r-optparse</requirement> </requirements> <command interpreter="Rscript" detect_errors="aggressive"> ##------------------------------------------------------------------------------------ ## This function reads the input file with the mapping between samples and files ## E.g. of result: ## mapping = { ## "e2t.ctab" : "sample1", ## "other.ctab" : "sample2", ## "i2t.ctab" : "sample1", ## "t_data.ctab": "sample1" ## ... ## } ##------------------------------------------------------------------------------------ #def read_sample_mapping_file(sample_mapping_file): #try #set mapping = {} #set file = open($sample_mapping_file.dataset.dataset.get_file_name(),'r') #for $line in $file: #set content= $line.strip().split('\t') #for $map in $content: #set mapping[$map]= $content[0] #end for #end for #return $mapping #except #return None #end try #end def ##------------------------------------------------------------------------------------ ## This function returns the name of the sample associated to a given file ##------------------------------------------------------------------------------------ #def get_sample_name($dataset, $sample_mapping): ##If the file with samples mapping was provided #if $sample_mapping != None: #return $sample_mapping.get($dataset.name, None) ##Otherwise with extract the sample name from the filename #else: #return str($dataset.element_identifier) #end if #end def ##------------------------------------------------------------------------------------ ## This function reads a dataset or list of datasets and sets the corresponding value ## in the $result variable ## e.g. of result ##'sample1' : { ## 'e_data': '/export/galaxy-central/database/files/000/dataset_13.dat' ## 'i_data': '/export/galaxy-central/database/files/000/dataset_10.dat', ## 't_data': '/export/galaxy-central/database/files/000/dataset_12.dat', ## 'e2t': '/export/galaxy-central/database/files/000/dataset_9.dat', ## 'i2t': '/export/galaxy-central/database/files/000/dataset_11.dat' ## }, ##------------------------------------------------------------------------------------ #def read_input_files($param_name, $param_value, $result, $sample_mapping, $create_if_empty): ## If input is a data collection #if isinstance($param_value, list): ## For each dataset #for $dataset in $param_value: ## Get the sample name #set sample_name = $get_sample_name($dataset, $sample_mapping) ## Check if sample is already registered #if not($result.has_key($sample_name)): #if ($create_if_empty == True): #set result[$sample_name] = {} #else: #raise ValueError("Error in input. Please check that input contains all the required files for sample " + $sample_name) #end if #end if ## Register the file to the sample #set result[$sample_name][$param_name] = str($dataset.dataset.dataset.get_file_name()) #end for #else: #if not($result.has_key("sample_1")): #set result["sample_1"] = {} #end if #set result["sample_1"][$param_name] = str($param_name.dataset.dataset.get_file_name()) #end if #return $result #end def ##------------------------------------------------------------------------------------ ## Main body of the tool ##------------------------------------------------------------------------------------ ## Set the params for the next R script #set result={} #set sample_mapping=None ## If the samples mapping file was provided, parse the content #if $samples_names != None and not(isinstance($samples_names, list) and (None in $samples_names)): #set sample_mapping = $read_sample_mapping_file($samples_names) #end if ## READ THE CONTENT FOR e_data AND STORE THE FILES ## INDEXED BY THEIR SAMPLE NAME ## e.g. 'HBR_Rep1' : { ## 'e_data': '/export/galaxy-central/database/files/000/dataset_13.dat' ## 'i_data': '/export/galaxy-central/database/files/000/dataset_10.dat', ## 't_data': '/export/galaxy-central/database/files/000/dataset_12.dat', ## 'e2t': '/export/galaxy-central/database/files/000/dataset_9.dat', ## 'i2t': '/export/galaxy-central/database/files/000/dataset_11.dat' ## }, ## 'HBR_Rep2' : {...} #set $result = $read_input_files("e_data.ctab", $e_data, $result, $sample_mapping, True) #set $result = $read_input_files("i_data.ctab", $i_data, $result, $sample_mapping, False) #set $result = $read_input_files("t_data.ctab", $t_data, $result, $sample_mapping, False) #set $result = $read_input_files("e2t.ctab", $e2t, $result, $sample_mapping, False) #set $result = $read_input_files("i2t.ctab", $i2t, $result, $sample_mapping, False) ## For each input sample, create a directory and link the input files for ballgown #import os #set n_sample = 1 #for $key, $value in $result.iteritems(): #set dir_name = str($output.files_path) + "/" + $key + "/" $os.makedirs($dir_name) #for $file_name, $file_path in $value.iteritems(): $os.symlink($file_path, $dir_name + $file_name) #end for #set n_sample = $n_sample + 1 #end for ## Run the R script with the location of the linked files and the name for outpot file ballgown.R --directory $output.files_path --outputtranscript $output --outputgenes $outputgn --texpression $trexpression --phendat $phendata --bgout $bgo </command> <inputs> <param name="e_data" type="data" multiple="true" format="tabular" label="Exon-level expression measurements" help="One row per exon. See below for more details."/> <param name="i_data" type="data" multiple="true" format="tabular" label="Intron- (i.e., junction-) level expression measurements" help="One row per intron. See below for more details."/> <param name="t_data" type="data" multiple="true" format="tabular" label="Transcript-level expression measurements" help="One row per transcript. See below for more details."/> <param name="e2t" type="data" multiple="true" format="tabular" label="Exons-transcripts mapping" help="Table with two columns, e_id and t_id, denoting which exons belong to which transcripts. See below for more details."/> <param name="i2t" type="data" multiple="true" format="tabular" label="Introns-transcripts mapping" help="Table with two columns, i_id and t_id, denoting which introns belong to which transcripts. See below for more details."/> <param name="samples_names" type="data" optional="true" multiple="false" format="tabular" label="File names for samples" help="Optional. Use in case that the names for the analysed samples cannot be extracted from the filenames."/> <param argument="--phendat" name="phendata" type="data" format="csv" label="phenotype data" /> <param argument="--texpression" name="trexpression" type="float" value="0.5" label="minimal transcript expression to appear in the results"/> </inputs> <outputs> <data name="bgo" format="rda" file="ballgown_object.rda" label="${tool.name} on ${on_string}: ballgown object (R data file)"/> <data name="output" format="csv" file="output_transcript.csv" label="${tool.name} on ${on_string}: transcripts expression (tabular)"/> <data name="outputgn" format="csv" file="output_genes.csv" label="${tool.name} on ${on_string}: genes expression (tabular)"/> </outputs> <tests> </tests> <help> ======================= Ballgown ======================= ----------------------- **What it does** ----------------------- Ballgown is a software package designed to facilitate flexible differential expression analysis of RNA-seq data. The Ballgown package provides functions to organize, visualize, and analyze the expression measurements for your transcriptome assembly. ---- ----------------------- **How to use** ----------------------- The input for this tools consists on 5 files for each sample in your experiment: - **e_data**: exon-level expression measurements. Tab file or collection of tab files. One row per exon. Columns are e_id (numeric exon id), chr, strand, start, end (genomic location of the exon), and the following expression measurements for each sample: * rcount: reads overlapping the exon * ucount: uniquely mapped reads overlapping the exon * mrcount: multi-map-corrected number of reads overlapping the exon * cov average per-base read coverage * cov_sd: standard deviation of per-base read coverage * mcov: multi-map-corrected average per-base read coverage * mcov_sd: standard deviation of multi-map-corrected per-base coverage - **i_data**: intron- (i.e., junction-) level expression measurements. Tab file or collection of tab files. One row per intron. Columns are i_id (numeric intron id), chr, strand, start, end (genomic location of the intron), and the following expression measurements for each sample: * rcount: number of reads supporting the intron * ucount: number of uniquely mapped reads supporting the intron * mrcount: multi-map-corrected number of reads supporting the intron - **t_data**: transcript-level expression measurements. Tab file or collection of tab files. One row per transcript. Columns are: * t_id: numeric transcript id * chr, strand, start, end: genomic location of the transcript * t_name: Cufflinks-generated transcript id * num_exons: number of exons comprising the transcript * length: transcript length, including both exons and introns * gene_id: gene the transcript belongs to * gene_name: HUGO gene name for the transcript, if known * cov: per-base coverage for the transcript (available for each sample) * FPKM: Cufflinks-estimated FPKM for the transcript (available for each sample) - **e2t**: Tab file or collection of tab files. Table with two columns, e_id and t_id, denoting which exons belong to which transcripts. These ids match the ids in the e_data and t_data tables. - **i2t**: Tab file or collection of tab files. Table with two columns, i_id and t_id, denoting which introns belong to which transcripts. These ids match the ids in the i_data and t_data tables. - samples_names: (optional) Tab file. Table with five columns, one row per sample. Defines which files from the input belong to each sample in the experiment. .. class:: infomark '''TIP''' *Note* Here's an example of a good phenotype data file for your expirement. +--------------+-------------------------+-------------------------+---+ |ids |experimental variable 1 |experimental variable 2 |...| +==============+=========================+=========================+===+ |sample 1 |value 1 |value 2 |...| +--------------+-------------------------+-------------------------+---+ |sample 2 |value 2 |value 1 |...| +--------------+-------------------------+-------------------------+---+ |sample 3 |value 1 |value 2 |...| +--------------+-------------------------+-------------------------+---+ |sample 4 |value 2 |value 1 |...| +--------------+-------------------------+-------------------------+---+ |... |value 1 |value 2 |...| +--------------+-------------------------+-------------------------+---+ .. class:: infomark *Note* The minimal transcript expression is a number used to filter the transcripts that are less or not expressed in our samples when compared to the genome ----------------------- **Outputs** ----------------------- This tool has 3 outputs: - **transcripts expression** : this is a csv file containing all the transcripts that are expressed above the transcripts expression value - **genes expression** : this is a csv file containing all the genes that are expressed above the transcripts expression value - **Ballgown object** : this is the ballgown object created during the process. This file can be re-used later for further analysis in a R console. ---- **Authors**: Théo Collard [SLU Global Bioinformatics Centre], Rafael Hernández de Diego [SLU Global Bioinformatics Centre], and Tomas Klingström [SLU Global Bioinformatics Centre] Sources are available at https://github.com/CollardT/Ballgown-Wrapper </help> </tool>