Merge pull request #149 from CCBR/rsem-isoform-matrix

feat: rsem-generate-data-matrix for genes & isoforms
CCBR · Aug 27, 2024 · 03e5dd9 · 03e5dd9
2 parents d26e6d1 + 827ca76
commit 03e5dd9
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 12 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -10,6 +10,7 @@
 ### Bug fixes
 
 - Ensure `renee build` creates necessary `config` directory during initialization. (#139, @kelly-sovacool)
+- Run `rsem-generate-data-matrix` to create gene and isoform matrix files. (#149, @kelly-sovacool)
 - Fix bug in the driver script that caused the snakemake module not to be loaded on biowulf in some cases. (#154, @kelly-sovacool)
 
 ### Documentation updates

diff --git a/workflow/Snakefile b/workflow/Snakefile
@@ -139,12 +139,8 @@ if paired_end:
             expand(join(workpath,degall_dir,"{name}.RSEM.isoforms.results"),name=samples),
             join(workpath,degall_dir,"RSEM.genes.FPKM.all_samples.txt"),
             join(workpath,degall_dir,"RSEM.isoforms.FPKM.all_samples.txt"),
-            #join(workpath,degall_dir,"RawCountFile_RSEM_genes_filtered.txt"),
-            #join(workpath,star_dir,"sampletable.txt"),
-
-            # PCA Reports
-            # expand(join(workpath,degall_dir,"PcaReport_{dtype}.html"),dtype=dtypes),
-
+            join(workpath, degall_dir, "RSEM.genes.expected_counts.all_samples.matrix"),
+            join(workpath, degall_dir, "RSEM.isoforms.expected_counts.all_samples.matrix"),
             # MultiQC
             join(workpath,"Reports","multiqc_report.html"),
 
@@ -202,11 +198,8 @@ elif not paired_end:
             expand(join(workpath,degall_dir,"{name}.RSEM.isoforms.results"),name=samples),
             join(workpath,degall_dir,"RSEM.genes.FPKM.all_samples.txt"),
             join(workpath,degall_dir,"RSEM.isoforms.FPKM.all_samples.txt"),
-            #join(workpath,degall_dir,"RawCountFile_RSEM_genes_filtered.txt"),
-            #join(workpath,star_dir,"sampletable.txt"),
-
-            # PCA Report
-            # expand(join(workpath,degall_dir,"PcaReport_{dtype}.html"),dtype=dtypes),
+            join(workpath, degall_dir, "RSEM.genes.expected_counts.all_samples.matrix"),
+            join(workpath, degall_dir, "RSEM.isoforms.expected_counts.all_samples.matrix"),
 
             # MultiQC
             join(workpath,"Reports","multiqc_report.html"),

diff --git a/workflow/rules/common.smk b/workflow/rules/common.smk
@@ -184,7 +184,7 @@ rule stats:
     """
 
 
-rule rsem_merge: # TODO is this redundant with `rsem-generate-data-matrix`? see https://github.com/CCBR/RENEE/issues/137
+rule rsem_merge:
     """Data processing step to merge the gene and isoform counts for each sample
     into count matrices.
     @Input:
@@ -213,6 +213,25 @@ rule rsem_merge: # TODO is this redundant with `rsem-generate-data-matrix`? see
         sed '1 s/^gene_id|GeneName/symbol/' > {output.reformatted}
     """
 
+rule rsem_data_matrix:
+    input:
+        genes=expand(join(workpath,degall_dir,"{name}.RSEM.genes.results"), name=samples),
+        isoforms=expand(join(workpath,degall_dir,"{name}.RSEM.isoforms.results"), name=samples),
+    output:
+        genes=join(workpath, degall_dir, "RSEM.genes.expected_counts.all_samples.matrix"),
+        isoforms=join(workpath, degall_dir, "RSEM.isoforms.expected_counts.all_samples.matrix")
+    params:
+        rname='pl:rsem_data_matrix',
+    envmodules:
+        config['bin'][pfamily]['tool_versions']['RSEMVER'],
+        config['bin'][pfamily]['tool_versions']['PYTHONVER'],
+    container: config['images']['rsem']
+    shell:
+        """
+        rsem-generate-data-matrix {input.genes} > {output.genes}
+        rsem-generate-data-matrix {input.isoforms} > {output.isoforms}
+        """
+
 
 rule rseqc:
     """