updated salmon transcriptome index generation

Snakefile

@@ -121,6 +121,7 @@ rule create_index_star:
         --sjdbGTFfile {input.gtf}) \
         1> {log.stdout} 2> {log.stderr}"
 
+
 rule extract_transcriptome:
     """ Create transcriptome from genome and gene annotations """
     input:
@@ -154,9 +155,49 @@ rule extract_transcriptome:
         -g {input.genome} {input.gtf}) \
         1> {log.stdout} 2> {log.stderr}"
 
-rule create_index_salmon:
+
+rule extract_decoys_salmon:
     """
-        Create index for Salmon quantification
+        Extract names of the genome targets
+    """
+    input:
+        genome = lambda wildcards:
+            samples_table['genome']
+            [samples_table['organism'] == wildcards.organism]
+            [0],
+
+    output:
+        decoys = os.path.join(
+                config['output_dir'],
+                "transcriptome",
+                "{organism}",
+                "decoys.txt")
+
+    singularity:
+        "docker://bash:5.0.16"
+
+    log:
+        stderr = os.path.join(
+            config['log_dir'],
+            "{organism}_extract_decoys_salmon.stderr.log"),
+        stdout = os.path.join(
+            config['log_dir'],
+            "{organism}_extract_decoys_salmon.stdout.log")
+
+    threads: 1
+
+    shell:
+        """
+        (grep "^>" <{input.genome} \
+        | cut -d " " -f 1 > {output.decoys} && \
+        sed -i.bak -e 's/>//g' {output.decoys}) \
+        1> {log.stdout} 2> {log.stderr}
+        """
+
+
+rule concatenate_transcriptome_and_genome:
+    """
+        Concatenate genome and transcriptome
     """
     input:
         transcriptome = os.path.join(
@@ -164,36 +205,78 @@ rule create_index_salmon:
                 "transcriptome",
                 "{organism}",
                 "transcriptome.fa",
+            ),
+        genome = lambda wildcards:
+            samples_table['genome']
+            [samples_table['organism'] == wildcards.organism]
+            [0],
+
+    output:
+        genome_transcriptome = os.path.join(
+                config['output_dir'],
+                "transcriptome",
+                "{organism}",
+                "genome_transcriptome.fa",
             )
+
+    singularity:
+        "docker://bash:5.0.16"
+
+    log:
+        stderr = os.path.join(
+            config['log_dir'],
+            "{organism}_concatenate_transcriptome_and_genome.stderr.log"),
+        stdout = os.path.join(
+            config['log_dir'],
+            "{organism}_concatenate_transcriptome_and_genome.stdout.log")
+
+    shell:
+        "(cat {input.transcriptome} {input.genome} \
+        1> {output.genome_transcriptome}) \
+        1> {log.stdout} 2> {log.stderr}"
+
+
+rule create_index_salmon:
+    """
+        Create index for Salmon quantification
+    """
+    input:
+        genome_transcriptome = os.path.join(
+                config['output_dir'],
+                "transcriptome",
+                "{organism}",
+                "genome_transcriptome.fa",
+            ),
+        decoys = os.path.join(
+                config['output_dir'],
+                "transcriptome",
+                "{organism}",
+                "decoys.txt")
     output:
         index = directory(
             os.path.join(
                 config['salmon_indexes'],
                 "{organism}",
-                "{kmer}",
                 "salmon.idx"))
 
-    params:
-        kmerLen = "{kmer}"
-
     singularity:
         "docker://zavolab/salmon:1.1.0-slim"
 
     log:
         stderr = os.path.join(
             config['log_dir'],
-            "{organism}_{kmer}_create_index_salmon.stderr.log"),
+            "{organism}_create_index_salmon.stderr.log"),
         stdout = os.path.join(
             config['log_dir'],
-            "{organism}_{kmer}_create_index_salmon.stdout.log")
+            "{organism}_create_index_salmon.stdout.log")
 
     threads: 8
 
     shell:
         "(salmon index \
-        --transcripts {input.transcriptome} \
+        --transcripts {input.genome_transcriptome} \
+        --decoys {input.decoys} \
         --index {output.index} \
-        --kmerLen {params.kmerLen} \
         --threads {threads}) \
         1> {log.stdout} 2> {log.stderr}"
 

tests/input_files/cluster.json

@@ -26,6 +26,18 @@
     "threads":"1",
     "mem":"1G"
   },
+  "extract_decoys_salmon":
+  {
+    "time": "00:30:00",
+    "threads":"1",
+    "mem":"10G"
+  },
+  "concatenate_transcriptome_and_genome":
+  {
+    "time": "00:30:00",
+    "threads":"1",
+    "mem":"10G"
+  },
   "create_index_salmon":
   {
     "time": "03:00:00",

tests/test_integration_workflow/expected_output.files

 results/kallisto_indexes/homo_sapiens/kallisto.idx
-results/salmon_indexes/homo_sapiens/31/salmon.idx/versionInfo.json
-results/salmon_indexes/homo_sapiens/31/salmon.idx/duplicate_clusters.tsv
-results/salmon_indexes/homo_sapiens/31/salmon.idx/info.json
+results/salmon_indexes/homo_sapiens/salmon.idx/versionInfo.json
+results/salmon_indexes/homo_sapiens/salmon.idx/duplicate_clusters.tsv
+results/salmon_indexes/homo_sapiens/salmon.idx/info.json
 results/star_indexes/homo_sapiens/75/STAR_index/chrLength.txt
 results/star_indexes/homo_sapiens/75/STAR_index/chrNameLength.txt
 results/star_indexes/homo_sapiens/75/STAR_index/chrName.txt

tests/test_integration_workflow/expected_output.md5

 cbaebdb67aee4784b64aff7fec9fda42  results/kallisto_indexes/homo_sapiens/kallisto.idx
-0ac1afd9a4f380afd70be75b21814c64  results/salmon_indexes/homo_sapiens/31/salmon.idx/versionInfo.json
-51b5292e3a874119c0e1aa566e95d70c  results/salmon_indexes/homo_sapiens/31/salmon.idx/duplicate_clusters.tsv
-4c1ab7841bbd1a1e8e3b15e7750ecc38  results/salmon_indexes/homo_sapiens/31/salmon.idx/info.json
+0ac1afd9a4f380afd70be75b21814c64  results/salmon_indexes/homo_sapiens/salmon.idx/versionInfo.json
+51b5292e3a874119c0e1aa566e95d70c  results/salmon_indexes/homo_sapiens/salmon.idx/duplicate_clusters.tsv
+92549565e504901d09216347f0b407f0  results/salmon_indexes/homo_sapiens/salmon.idx/info.json
 dee7cdc194d5d0617552b7a3b5ad8dfb  results/star_indexes/homo_sapiens/75/STAR_index/chrLength.txt
 8e2e96e2d6b7f29940ad5de40662b7cb  results/star_indexes/homo_sapiens/75/STAR_index/chrNameLength.txt
 d0826904b8afa45352906ad9591f2bfb  results/star_indexes/homo_sapiens/75/STAR_index/chrName.txt

workflow/rules/paired_end.snakefile.smk

@@ -303,7 +303,6 @@ rule pe_quantification_salmon:
             os.path.join(
                 config["salmon_indexes"],
                 str(samples_table.loc[wildcards.sample, "organism"]),
-                str(samples_table.loc[wildcards.sample, "kmer"]),
                 "salmon.idx")
 
     output:

workflow/rules/single_end.snakefile.smk

@@ -247,7 +247,6 @@ rule quantification_salmon:
             os.path.join(
                 config["salmon_indexes"],
                 str(samples_table.loc[wildcards.sample, "organism"]),
-                str(samples_table.loc[wildcards.sample, "kmer"]),
                 "salmon.idx"),
         gtf = lambda wildcards:
             samples_table.loc[wildcards.sample, "gtf_filtered"]