Preseq now actually reads the BAM

README.md

@@ -37,7 +37,7 @@ The following dependencies are downloaded when running with `--use-conda`, other
   + `samblaster`
   + `parallel` (preferably version 20201122)
   + `bedtools`
-  + `preseq` (version 3.1.2+)
+  + `preseq` (version 3.1.2+, must be compiled with htslib enabled)
   + `fastqc`
   + `trim_galore`
   + `fastq_screen` (only required if running `fastq_screen`)

config/config.yaml

@@ -85,7 +85,9 @@ fastq_screen_conf: /primary/projects/bbc/references/fastq_screen/downloaded_2020
 # Run preseq on data
 # If True - run preseq
 # If False - skip this step
+# preseq_options - other options to include in preseq c_curve (output file, paired end, and BAM input options already included)
 preseq: False
+preseq_options: ''
 
 # Calculate the ratio observed/expected coverage based on bismap results for different genomic features
 # run - create obs/exp plots if True, skip if False

workflow/rules/qc.smk

@@ -73,6 +73,7 @@ rule preseq:
     params:
         dir = f'{output_directory}/analysis/preseq',
         out = f'{output_directory}/analysis/preseq/{{sample}}.ccurve.txt',
+        opt = config['preseq_options'],
     log:
         f'{output_directory}/logs/preseq/{{sample}}.log',
     benchmark:
@@ -88,7 +89,7 @@ rule preseq:
     shell:
         """
         mkdir -p {params.dir}
-        preseq c_curve -o {params.out} -P {input.bam} 2> {log}
+        preseq c_curve {params.opt} -o {params.out} -P -B {input.bam} 2> {log}
         """
 
 def get_multiQC_params(wildcards):