Decoupler Pathway Inference

tools/tertiary-analysis/decoupler/decoupler_pathway_inference.py

@@ -0,0 +1,105 @@
+# import the necessary packages
+import argparse
+
+import anndata as ad
+import decoupler as dc
+import pandas as pd
+import hdf5plugin
+
+# define arguments for the script
+parser = argparse.ArgumentParser()
+
+# add AnnData input file option
+parser.add_argument(
+    "-i", "--input_anndata", help="AnnData input file", required=True
+)
+
+# add network input file option
+parser.add_argument(
+    "-n", "--input_network", help="Network input file", required=True
+)
+
+# output file prefix
+parser.add_argument(
+    "--output",
+    help="output files prefix",
+    default=None,
+)
+
+# figure size option
+parser.add_argument(
+    "-f", "--figure_size", help="figure size", default="10,10"
+)
+
+# path to save Activities AnnData file
+parser.add_argument(
+    "-a", "--activities_path", help="Path to save Activities AnnData file", default=None
+)
+
+# Column name in net with source nodes
+parser.add_argument(
+    "-s", "--source", help="Column name in net with source nodes.", default="source"
+)
+
+# Column name in net with target nodes
+parser.add_argument(
+    "-t", "--target", help="Column name in net with target nodes.", default="target"
+)
+
+# Column name in net with weights.
+parser.add_argument(
+    "-w", "--weight", help="Column name in net with weights.", default="weight"
+)
+
+# add boolean argument for use_raw
+parser.add_argument(
+    "--use_raw", action="store_true", default=False, help="Whether to use the raw part of the AnnData object"
+)
+
+# add argument for min_cells
+parser.add_argument(
+    "--min_n", help="Minimum of targets per source. If less, sources are removed.", default=5, type=int
+)
+args = parser.parse_args()
+
+# check that either -o or --output is specified
+if args.output is None:
+    raise ValueError("Please specify either -o or --output")
+
+# read in the AnnData input file
+adata = ad.read_h5ad(args.input_anndata)
+
+# read in the input file network input file
+network = pd.read_csv(args.input_network, sep='\t')
+
+if (
+    args.source not in network.columns
+    or args.target not in network.columns
+    or args.weight not in network.columns
+):
+    raise ValueError(
+        "Source, target, and weight columns are not present in the network"
+    )
+
+
+print(type(args.min_n))
+dc.run_mlm(
+    mat=adata,
+    net=network,
+    source=args.source,
+    target=args.target,
+    weight=args.weight,
+    verbose=True,
+    min_n=args.min_n,
+    use_raw=args.use_raw #Failing at the moment
+)
+
+if args.output is not None:
+    # write adata.obsm[ulm_key] and adata.obsm[ulm_pvals_key] to the output network files
+    adata.obsm["mlm_estimate"].to_csv(args.output + "_mlm.tsv", sep="\t")
+    adata.obsm["mlm_pvals"].to_csv(args.output + "_mlm_pvals.tsv", sep="\t")
+
+# if args.activities_path is specified, generate the activities AnnData and save the AnnData object to the specified path
+if args.activities_path is not None:
+    acts = dc.get_acts(adata, obsm_key="mlm_estimate")
+    adata.write_h5ad(args.activities_path)

tools/tertiary-analysis/decoupler/decoupler_pathway_inference.xml

@@ -0,0 +1,90 @@
+<tool id="decoupler_pathway_inference" name="Decoupler Pathway Inference" version="1.4.0+galaxy0" profile="20.05" license="MIT">
+    <description>
+        Pathway inference using the Decoupler.
+    </description>
+    <requirements>
+        <requirement type="package" version="1.4.0">decoupler</requirement>
+    </requirements>
+    <command>
+        python '$__tool_directory__/decoupler_pathway_inference.py'
+            --input_file '$input_file'
+            --input_network '$input_network_file'
+            --min_n '$min_n'
+            '$use_raw'
+            --output "inference"
+            --activities_path anndata_activities_path.h5ad
+    </command>
+    <inputs>
+        <param name="input_anndata" type="data" format="h5ad" label="Input AnnData file" />
+        <param name="input_network_file" type="data" format="tabular" label="Input Network file" />
+        <param name="min_n" type="integer" min="0" value="5" label="Minimum of targets per source. If less, sources are removed" />
+        <param name="use_raw" type="boolean" truevalue="--use_raw" falsevalue="" checked="false" label="Whether to use the raw part of the AnnData object" />
+        <param name="write_inference" type="boolean" truevalue="--write_inference" falsevalue="" checked="true" label="Write the inference TSV files"  />
+        <param name="write_activities_path" type="boolean" truevalue="--write_activities_path" falsevalue="" checked="true" label="Write the modified AnnData object" />
+    </inputs>
+    <outputs>
+        <data name="output_ad" format="h5ad" from_work_dir="anndata_activities_path.h5ad" label="${tool.name} on ${on_string}: Output AnnData file" />
+        <data name="output_table_mlm_estimate" format="tabular" from_work_dir="inference_mlm.tsv" label="${tool.name} on ${on_string}: Output MLM estimate table" />
+        <data name="output_table_mlm_pvalue" format="tabular" from_work_dir="inference_mlm_pvals.tsv" label="${tool.name} on ${on_string}: Output MLM p-values table" />
+    </outputs>
+    <tests>
+        <!-- Hint: You can use [ctrl+alt+t] after defining the inputs/outputs to auto-scaffold some basic test cases. -->
+
+    <test expect_num_outputs="3">
+        <param name="input_anndata" value="pbmc3k_processed.h5ad"/>
+        <param name="input_network_file" value="progeny_test.tsv"/>
+        <param name="min_n" value="5"/>
+        <param name="use_raw" value="false"/>
+        <param name="write_inference" value="true"/>
+        <param name="write_activities_path" value="true"/>
+        <output name="output_ad">
+            <assert_contents>
+                <has_h5_keys keys="obs/mlm_estimate"/>
+            </assert_contents>
+        </output>
+        <output name="output_table_mlm_estimate">
+            <assert_contents>
+                <has_n_columns n="5"/>
+            </assert_contents>
+        </output>
+        <output name="output_table_mlm_pvalue">
+            <assert_contents>
+                <has_n_columns n="5"/>
+            </assert_contents>
+        </output>
+    </test>
+    </tests>
+    <help>
+**What it does**
+
+Usage
+.....
+
+
+**Description**
+
+This tool scores cells using the AUCell method for gene sets.
+
+**Input**
+
+The input file should be an AnnData object in H5AD format. The tool accepts an H5AD file containing raw or normalized data.
+
+The tool also takes network file containing a collection of pathways and their target genes, with weights for each interaction
+
+You can also specify whether to use the raw data in the AnnData object instead of the X matrix using the "use_raw" parameter and Minimum of targets per source using "min_n".
+
+
+**Output**
+
+The tool outputs an AnnData object containing the scores in the "obs" field, and tab-separated text files containing the scores for each cell.
+
+If the "write_activities_path" parameter is set to "true", the tool will write the modified AnnData object to an H5AD file. 
+If the "write_inference" parameter is set to "true", the tool will output a tab-separated text file containing the scores for each cell.
+
+
+
+    </help>
+    <citations>
+        <citation type="doi">10.1093/bioadv/vbac016 </citation>
+    </citations>
+</tool>

tools/tertiary-analysis/decoupler/get_test_data.sh

@@ -19,3 +19,10 @@ function get_data {
 mkdir -p test-data
 pushd test-data
 get_data $MTX_LINK $BASENAME_FILE
+
+
+BASENAME_FILE='pbmc3k_processed.h5ad'
+
+MTX_LINK='https://zenodo.org/records/3752813/files/pbmc3k_processed.h5ad'
+
+get_data $MTX_LINK $BASENAME_FILE