From 0997adf59b9e631e1b61a62b48419e3adde03320 Mon Sep 17 00:00:00 2001
From: Brett Copeland <brcopeland@gmail.com>
Date: Wed, 23 Jun 2021 01:32:34 -0700
Subject: [PATCH] Added wrapper for rsem-generate-data-matrix. (#374)

* Added wrapper for rsem-generate-data-matrix.

* Added a second results file to demonstrate merging.
---
 bio/rsem/generate-data-matrix/environment.yaml  |  4 ++++
 bio/rsem/generate-data-matrix/meta.yaml         | 11 +++++++++++
 bio/rsem/generate-data-matrix/test/Snakefile    | 14 ++++++++++++++
 .../generate-data-matrix/test/a.genes.results   |  2 ++
 .../generate-data-matrix/test/b.genes.results   |  2 ++
 bio/rsem/generate-data-matrix/wrapper.py        | 17 +++++++++++++++++
 test.py                                         |  8 ++++++++
 7 files changed, 58 insertions(+)
 create mode 100644 bio/rsem/generate-data-matrix/environment.yaml
 create mode 100644 bio/rsem/generate-data-matrix/meta.yaml
 create mode 100644 bio/rsem/generate-data-matrix/test/Snakefile
 create mode 100644 bio/rsem/generate-data-matrix/test/a.genes.results
 create mode 100644 bio/rsem/generate-data-matrix/test/b.genes.results
 create mode 100644 bio/rsem/generate-data-matrix/wrapper.py

diff --git a/bio/rsem/generate-data-matrix/environment.yaml b/bio/rsem/generate-data-matrix/environment.yaml
new file mode 100644
index 00000000000..2637f709fbc
--- /dev/null
+++ b/bio/rsem/generate-data-matrix/environment.yaml
@@ -0,0 +1,4 @@
+channels:
+  - bioconda
+dependencies:
+  - rsem ==1.3.3
diff --git a/bio/rsem/generate-data-matrix/meta.yaml b/bio/rsem/generate-data-matrix/meta.yaml
new file mode 100644
index 00000000000..9780957febb
--- /dev/null
+++ b/bio/rsem/generate-data-matrix/meta.yaml
@@ -0,0 +1,11 @@
+name: rsem generate data matrix
+description: |
+        Run rsem-generate-data-matrix to combine a set of single-sample rsem results into a single matrix.
+authors:
+  - Brett Copeland
+input:
+  - a list of rsem results files
+output:
+  - Quantification results summarized by allele/gene/isoform per sample
+notes: |
+  * For more information, see https://github.com/deweylab/RSEM.
diff --git a/bio/rsem/generate-data-matrix/test/Snakefile b/bio/rsem/generate-data-matrix/test/Snakefile
new file mode 100644
index 00000000000..1701cd1ab8d
--- /dev/null
+++ b/bio/rsem/generate-data-matrix/test/Snakefile
@@ -0,0 +1,14 @@
+rule rsem_generate_data_matrix:
+    input:
+        # one or more expression files created by rsem-calculate-expression
+        ["a.genes.results", "b.genes.results"],
+    output:
+        # a tsv containing each sample in the input as a column
+        "genes.results",
+    params:
+        # optional additional parameters
+        extra="",
+    log:
+        "logs/rsem/generate_data_matrix.log",
+    wrapper:
+        "master/bio/rsem/generate-data-matrix"
diff --git a/bio/rsem/generate-data-matrix/test/a.genes.results b/bio/rsem/generate-data-matrix/test/a.genes.results
new file mode 100644
index 00000000000..b9772cd16d1
--- /dev/null
+++ b/bio/rsem/generate-data-matrix/test/a.genes.results
@@ -0,0 +1,2 @@
+gene_id	transcript_id(s)	length	effective_length	expected_count	TPM	FPKM
+Sheila	Sheila	20.00	20.00	0.00	0.00	0.00
diff --git a/bio/rsem/generate-data-matrix/test/b.genes.results b/bio/rsem/generate-data-matrix/test/b.genes.results
new file mode 100644
index 00000000000..ebdf3046fec
--- /dev/null
+++ b/bio/rsem/generate-data-matrix/test/b.genes.results
@@ -0,0 +1,2 @@
+gene_id	transcript_id(s)	length	effective_length	expected_count	TPM	FPKM
+Sheila	Sheila	20.00	20.00	1.00	1.00	1.00
diff --git a/bio/rsem/generate-data-matrix/wrapper.py b/bio/rsem/generate-data-matrix/wrapper.py
new file mode 100644
index 00000000000..f6f5a09299e
--- /dev/null
+++ b/bio/rsem/generate-data-matrix/wrapper.py
@@ -0,0 +1,17 @@
+__author__ = "Brett Copeland"
+__copyright__ = "Copyright 2021, Brett Copeland"
+__email__ = "brcopeland@ucsd.edu"
+__license__ = "MIT"
+
+
+import os
+
+from snakemake.shell import shell
+
+extra = snakemake.params.get("extra", "")
+log = snakemake.log_fmt_shell(stdout=False, stderr=True)
+shell(
+    "rsem-generate-data-matrix {extra} "
+    "{snakemake.input} > {snakemake.output} "
+    "{log}"
+)
diff --git a/test.py b/test.py
index 6cf757787d5..674b097af6f 100644
--- a/test.py
+++ b/test.py
@@ -3935,3 +3935,11 @@ def test_rsem_prepare_reference():
         "bio/rsem/prepare-reference",
         ["snakemake", "--cores", "1", "--use-conda", "-F"],
     )
+
+
+@skip_if_not_modified
+def test_generate_data_matrix():
+    run(
+        "bio/rsem/generate-data-matrix",
+        ["snakemake", "--cores", "1", "--use-conda", "-F"],
+    )