attempt to fix the download of gene gtf

ATAC-seq_pipeline/Snakefile

@@ -16,6 +16,7 @@ WORKING_DIR         = config["working_dir"]
 RESULT_DIR          = config["results_dir"]
 DATA_DIR            = config["data_dir"]
 GENOME_FASTA_URL    = config["genome_fasta_url"]
+GENE_GTF_URL        = config["gene_gtf_url"]
 
 #units = pd.read_table(config["units"], dtype=str).set_index(["bed"], drop=False)
 

ATAC-seq_pipeline/config.yaml

@@ -11,6 +11,7 @@ units : "units.tsv"
 ## genome reference ##
 
 genome_fasta_url : "ftp://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_mouse/release_M17/GRCm38.primary_assembly.genome.fa.gz"
+gene_gtf_url: "ftp://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_mouse/release_M25/gencode.vM25.annotation.gtf.gz"
 HMMRATAC: "https://github.com/LiuLabUB/HMMRATAC/releases/download/1.2.10/HMMRATAC_V1.2.10_exe.jar"
 
 
@@ -55,4 +56,4 @@ bamCoverage:
 
 # computeMatrix
 ComputeMatrix:
-  GTF: "gencode.vM25.annotation.gtf"  
+  GTF: "gencode.vM25.annotation.gtf"

ATAC-seq_pipeline/rules/deeptools.smk

@@ -71,7 +71,7 @@ rule ComputeMatrix:
     conda:
         "envs/deeptools.yaml"
     params:
-        GTF = str(config['ComputeMatrix']['GTF'])
+        GTF = WORKING_DIR + "gtf_gene.gtf"
     shell:
         "computeMatrix reference-point -S {input} -R {params.GTF} -o {output} -a 3000 -b 3000 2>{log}"
 

ATAC-seq_pipeline/rules/external_data.smk

@@ -9,6 +9,12 @@ rule download_genome:
         "wget -O {output} {GENOME_FASTA_URL}"
 
 
+rule download_gene_gtf:
+    output:
+        WORKING_DIR + "gtf_gene.gtf"
+    shell:
+        "curl {GENE_GTF_URL} | gunzip -c > {output}"
+
 rule index_genome:
     input:
         WORKING_DIR + "reference",