replace dedupe.sh with clumpify.sh for duplicates removal

rules/preprocess.smk

@@ -216,32 +216,37 @@ rule dedupe:
         rev_clean = OUT_DIR+'/{dataset}/preQC/bfiltered/{fastq_file}-R2.clean.fastq.gz',
     output:
         fwd_clean_dedup = temp(OUT_DIR+'/{dataset}/preQC/cdedupe/{fastq_file}-R1.clean.nodup.fastq.gz'),
-        rev_clean_dedup = temp(OUT_DIR+'/{dataset}/preQC/cdedupe/{fastq_file}-R2.clean.nodup.fastq.gz'),
-        fastq_duplicates = OUT_DIR+'/{dataset}/preQC/cdedupe/{fastq_file}.clean.duplicates.fastq.gz'
+        rev_clean_dedup = temp(OUT_DIR+'/{dataset}/preQC/cdedupe/{fastq_file}-R2.clean.nodup.fastq.gz')
     log:
         OUT_DIR + '/{dataset}/preQC/logs/cdedupe/{fastq_file}.log'
     threads:cpus_avail
     params:
-        dd_mem_gb = (config['preprocess']['filter_human']['bbmap_mem']/3)*2,
-        rf_mem_gb = config['preprocess']['filter_human']['bbmap_mem']/4
+        dd_mem_gb = config['preprocess']['filter_human']['bbmap_mem']
     singularity: singularity_img
     group: 'preprocess'
     message: "Running dedupe with {threads} cores."
     shell:
         '''
-        (dedupe.sh \
+        (clumpify.sh \
         in1={input.fwd_clean} in2={input.rev_clean} \
-        out=stdout.fq \
-        outd={output.fastq_duplicates} \
-        ac=f minidentity=99 \
-        -Xmx{params.dd_mem_gb}g| \
-        reformat.sh \
-        int=t in=stdin.fq \
         out1={output.fwd_clean_dedup} \
         out2={output.rev_clean_dedup} \
-        threads={threads} \
-        -Xmx{params.rf_mem_gb}g) &>{log}
+        dedupe=t \
+        t={threads} \
+        -Xmx{params.dd_mem_gb}g -eoom) &>{log}
         '''
+        #(dedupe.sh \
+        #in1={input.fwd_clean} in2={input.rev_clean} \
+        #out=stdout.fq \
+        #outd={output.fastq_duplicates} \
+        #ac=f minidentity=99 \
+        #-Xmx{params.dd_mem_gb}g| \
+        #reformat.sh \
+        #int=t in=stdin.fq \
+        #out1={output.fwd_clean_dedup} \
+        #out2={output.rev_clean_dedup} \
+        #threads={threads}) &>{log}
+        #'''
 # After removal of adapters, human reads, and duplicates,
 # the reads' 3'end are quality trimmed (cut by quality score) with fastp
 # Notice that adapter- and quality- filtering are disabled because it was done by rule trim_adapters