update

dunnart/Snakefile

@@ -6,12 +6,11 @@
     # 3. Filtering
     # 4. Alignment QC & Library Complexity
     # 5. deepTools
-    # 6. phantomPeakQuals
+    # 6. cross correlation (SPP)
     # 7. Call narrow peaks (MACS2)
     # 8. Peak QC
     # 9. Create consensus peaksets
-    # 10. Annotate peaks relative to gene features (HOMER)
-    # 11. Present QC for raw read, alignment, peak-calling in MultiQC
+    # 10. Present QC for raw read, alignment, peak-calling in MultiQC
 
 
 configfile: "envs/config.yaml"
@@ -82,7 +81,6 @@ rule all:
         "results/qc/bamPEFragmentSize_rawcounts.tab",
         expand("results/bowtie2/{sample}_R1_trimmed_q30.bam", sample=all_samples),
         expand("logs/{sample}_filt_15Mreads.SE.spp.log", sample=all_samples),
-        # expand("results/qc/{sample}_readlength.txt", sample=all_samples),
         expand("results/qc/{sample}_filt_15Mreads.SE.cc.qc", sample=all_samples),
         expand("results/qc/{sample}_filt_15Mreads.SE.cc.plot.pdf", sample=all_samples),
         expand("results/macs2/{case}_vs_{control}_macs2_peaks.narrowPeak", zip, case=IPS, control=INPUTS),
@@ -460,8 +458,7 @@ rule subsample_aligned_reads:
         shell("zcat {input} > {output.tmp}")
 
 # Determine exclusion range for fragment length estimation.
-## From bamPEFragmentSize (deepTools) average fragment length is ~220bp
-## See bamPEFragmentSize.histogram.png
+
 # Use a fixed lowerbound at -500.
 # Upperbound E
 #EXCLUSION_RANGE_MAX is
@@ -479,7 +476,7 @@ rule cross_correlation_SSP:
         "logs/{sample}_filt_15Mreads.SE.spp.log"
     params:
         EXCLUSION_RANGE_MIN=-500,
-        EXCLUSION_RANGE_MAX=230
+        EXCLUSION_RANGE_MAX=60
     shell:
         "Rscript scripts/run_spp.R -c={input} -savp={output.CC_PLOT_FILE} -out={output.CC_SCORES_FILE} -x={params.EXCLUSION_RANGE_MIN}:{params.EXCLUSION_RANGE_MAX} 2> {log}"
 
@@ -509,10 +506,10 @@ rule call_peaks_macs2:
         name = "{case}_vs_{control}_macs2",
     shell:
         " macs2 callpeak -f BAMPE -t {input.case} \
-        -c {input.control} \
+        -c {input.control} --keep-dup all \
         --outdir results/macs2/ \
         -n {params.name} \
-        -g 2740338543 -B 2> {log} "
+        -g 2740338543 2> {log} "
 
 # ===============================================================================================
 #  8. Peak QC