frip plot working

Author: niekwit

workflow/rules/bed.smk

@@ -184,7 +184,6 @@ if config["peak_calling_perl"]["run"]:
             "-b stdin | "
             "awk '{{i+=$NF}}END{{print i}}' > "
             "{output.peak_read_count} "
-            "{log}"
 
 
     rule plot_fraction_of_reads_in_peaks:

workflow/rules/peak_calling.smk

@@ -1,5 +1,3 @@
-from snakemake.logging import logger
-
 if config["peak_calling_perl"]["run"]:
     fdr = config["peak_calling_perl"]["fdr"]
     rule peak_calling_perl:
@@ -71,7 +69,7 @@ if config["peak_calling_macs2"]["run"]:
             resources:
                 runtime=config["resources"]["deeptools"]["time"]
             log:
-                "logs/consensus_peaks_macs2_narrow_fdr{fdr}/{bg_sample}.log"
+                "logs/consensus_peaks_macs2_narrow/fdr{fdr}/{bg_sample}.log"
             conda:
                 "../envs/peak_calling.yaml"
             shell:
@@ -113,7 +111,7 @@ if config["peak_calling_macs2"]["run"]:
             resources:
                 runtime=config["resources"]["plotting"]["time"]
             log:
-                "logs/plotting/macs2_narrow_peak_annotation_plots_fdr{fdr}.log"
+                "logs/plotting/macs2_narrow_peak_annotation_plots/fdr{fdr}.log"
             conda:
                 "../envs/R.yaml"
             script:
@@ -130,7 +128,7 @@ if config["peak_calling_macs2"]["run"]:
             params:
                 extra=""
             log:
-                "logs/annotate_peaks_macs2_narrow_fdr{fdr}/{bg_sample}.log"
+                "logs/annotate_peaks_macs2_narrow/fdr{fdr}/{bg_sample}.log"
             threads: config["resources"]["deeptools"]["cpu"]
             resources:
                 runtime=config["resources"]["deeptools"]["time"]
@@ -293,7 +291,7 @@ if config["peak_calling_macs2"]["run"]:
             resources:
                 runtime=config["resources"]["deeptools"]["time"]
             log:
-                "logs/consensus_peaks_macs2_broad_fdr{fdr}/{bg_sample}.log"
+                "logs/consensus_peaks_macs2_broad/fdr{fdr}/{bg_sample}.log"
             conda:
                 "../envs/peak_calling.yaml"
             shell:
@@ -421,55 +419,55 @@ if config["peak_calling_macs2"]["run"]:
                     "../scripts/plot_enrichment.R"
 
 
-            rule count_reads_in_peaks:
-            # Adapted from https://www.biostars.org/p/337872/#337890
-                input:
-                    bam="results/bam/{dir}/{bg_sample}.bam",
-                    b="results/macs2_broad/fdr{fdr}/{dir}/{bg_sample}_peaks.broadPeak",
-                output:
-                    total_read_count="results/macs2_broad/fdr{fdr}/read_counts/{dir}/{bg_sample}.total.count",
-                    peak_read_count="results/macs2_broad/fdr{fdr}/read_counts/{dir}/{bg_sample}.peak.count",
-                params:
-                    extra="",
-                threads: config["resources"]["deeptools"]["cpu"]
-                resources:
-                    runtime=config["resources"]["deeptools"]["time"]
-                log:
-                    "logs/bedtools_intersect/fdr{fdr}/{dir}/{bg_sample}.log"
-                conda:
-                    "../envs/peak_calling.yaml"
-                shell:
-                    "bedtools bamtobed "
-                    "{params.extra} "
-                    "-i {input.bam} | "
-                    "sort -k1,1 -k2,2n | "
-                    "tee >(wc -l > {output.total_read_count}) | "
-                    "bedtools intersect "
-                    "{params.extra} "
-                    "-sorted "
-                    "-c "
-                    "-a {input.b} "
-                    "-b stdin | "
-                    "awk '{{i+=$NF}}END{{print i}}' > "
-                    "{output.peak_read_count} "
-                    "{log}"
-                
+        rule count_reads_in_peaks:
+        # Adapted from https://www.biostars.org/p/337872/#337890
+            input:
+                bam="results/bam/{dir}/{bg_sample}.bam",
+                b="results/macs2_broad/fdr{fdr}/{dir}/{bg_sample}_peaks.broadPeak",
+            output:
+                total_read_count="results/macs2_broad/fdr{fdr}/read_counts/{dir}/{bg_sample}.total.count",
+                peak_read_count="results/macs2_broad/fdr{fdr}/read_counts/{dir}/{bg_sample}.peak.count",
+            params:
+                extra="",
+            threads: config["resources"]["deeptools"]["cpu"]
+            resources:
+                runtime=config["resources"]["deeptools"]["time"]
+            log:
+                "logs/bedtools_intersect/fdr{fdr}/{dir}/{bg_sample}.log"
+            conda:
+                "../envs/peak_calling.yaml"
+            shell:
+                "bedtools bamtobed "
+                "{params.extra} "
+                "-i {input.bam} | "
+                "sort -k1,1 -k2,2n | "
+                "tee >(wc -l > {output.total_read_count}) | "
+                "bedtools intersect "
+                "{params.extra} "
+                "-sorted "
+                "-c "
+                "-a {input.b} "
+                "-b stdin | "
+                "awk '{{i+=$NF}}END{{print i}}' > "
+                "{output.peak_read_count} "
+                #"{log}"
+            
 
-            rule plot_fraction_of_reads_in_peaks:
-                input:
-                    total_read_count=expand("results/macs2_broad/fdr{fdr}/read_counts/{dir}/{bg_sample}.total.count", dir=DIRS, fdr=fdr, bg_sample=BG_SAMPLES),
-                    peak_read_count=expand("results/macs2_broad/fdr{fdr}/read_counts/{dir}/{bg_sample}.peak.count", dir=DIRS, fdr=fdr, bg_sample=BG_SAMPLES),
-                output:
-                    plot="results/plots/macs2_broad/fdr{fdr}/frip.pdf",
-                    csv="results/macs2_broad/fdr{fdr}/frip.csv",
-                params:
-                    extra="",
-                threads: config["resources"]["plotting"]["cpu"]
-                resources:
-                    runtime=config["resources"]["plotting"]["time"]
-                log:
-                    "logs/plot_frip/fdr{fdr}.log"
-                conda:
-                    "../envs/R.yaml"
-                script:
-                    "../scripts/plot_frip.R"
+        rule plot_fraction_of_reads_in_peaks:
+            input:
+                total_read_count=expand("results/macs2_broad/fdr{fdr}/read_counts/{dir}/{bg_sample}.total.count", dir=DIRS, fdr=fdr, bg_sample=BG_SAMPLES),
+                peak_read_count=expand("results/macs2_broad/fdr{fdr}/read_counts/{dir}/{bg_sample}.peak.count", dir=DIRS, fdr=fdr, bg_sample=BG_SAMPLES),
+            output:
+                plot="results/plots/macs2_broad/fdr{fdr}/frip.pdf",
+                csv="results/macs2_broad/fdr{fdr}/frip.csv",
+            params:
+                extra="",
+            threads: config["resources"]["plotting"]["cpu"]
+            resources:
+                runtime=config["resources"]["plotting"]["time"]
+            log:
+                "logs/plot_frip/fdr{fdr}.log"
+            conda:
+                "../envs/R.yaml"
+            script:
+                "../scripts/plot_frip.R"

workflow/scripts/general_functions.smk

@@ -34,7 +34,7 @@ def targets():
             expand("results/peaks/fdr{fdr}/consensus_peaks/{bg_sample}.annotated.txt", fdr=fdr, bg_sample=BG_SAMPLES),
             expand("results/peaks/fdr{fdr}/consensus_peaks/{bg_sample}.geneIDs.txt", fdr=fdr,bg_sample=BG_SAMPLES),
             expand("results/plots/peaks/fdr{fdr}/frip.pdf", fdr=fdr),
-            expand("results/plots/peaks/fdr{fdr}/frip.csv", fdr=fdr),
+            expand("results/peaks/fdr{fdr}/frip.csv", fdr=fdr),
             ])
         if config["consensus_peaks"]["enrichment_analysis"]["run"]:
             TARGETS.extend([

workflow/scripts/plot_frip.R

@@ -7,18 +7,19 @@ library(tidyverse)
 library(cowplot)
 
 # Plot bar graph of FRiP values
-total_read_counts <- snakemake@input[["total_read_counts"]]
+total_read_counts <- snakemake@input[["total_read_count"]]
 peak_read_counts <- snakemake@input[["peak_read_count"]]
 
-# df to store FRiP values
+# Create df to store FRiP values
 df <- data.frame(sample = character(),
                  read_in_peaks = numeric(),
                  total_reads = numeric(),
                  fraction_of_reads_in_peaks = numeric())
 
 # Add FRiP values to df
 for (i in seq_along(total_read_counts)) {
-  sample <- snakemake@wildcards[["bg_sample"]][i]
+  sample <- basename(total_read_counts[i]) %>% str_remove(".total.count")
+  sample <- paste(basename(dirname(total_read_counts[i])), sample, sep="_")
   total_reads <- as.numeric(read_lines(total_read_counts[i]))
   peak_reads <- as.numeric(read_lines(peak_read_counts[i]))
   frip <- peak_reads / total_reads
@@ -30,6 +31,9 @@ for (i in seq_along(total_read_counts)) {
             fraction_of_reads_in_peaks = frip)
 }
 
+# Save df to file
+write.csv(df, snakemake@output[["csv"]], row.names = FALSE)
+
 # Create plot
 p <- ggplot(df, aes(x = sample, 
                     y = fraction_of_reads_in_peaks)) +
@@ -43,11 +47,10 @@ p <- ggplot(df, aes(x = sample,
                      limits = c(0, 1)) +
   labs(title = NULL,
        x = NULL,
-       y = "Fraction of reads in peaks") +
-  theme(plot.title = element_text(hjust = 0.5))
+       y = "Fraction of reads in peaks")
 
 # Save plot
-ggsave(snakemake@output[[1]], p)
+ggsave(snakemake@output[["plot"]], p)
 
 # Close redirection of output/messages
 sink(log, type = "output")