feat(rule): add alignment summary

Author: fcaretti

Diff for: workflow/Snakefile

@@ -20,19 +20,20 @@ known_sites_idx = f"{known_sites}.idx"
 sample_files = glob.glob(os.path.join(data_folder, "*.bam"))
 samples = [os.path.basename(f).replace(".bam", "") for f in sample_files]
 
-read_groups = [f"results/grouped/{sample}.bam" for sample in samples]
-deduped_files = [f"results/dedup/{sample}.bam" for sample in samples]
-splitted_files = [f"results/split/{sample}.bam" for sample in samples]
 recalibrated_files = [f"results/recal/{sample}.bam" for sample in samples]
 
-
+first_summaries = [f"results/stats/{sample}_before_recal.summary.txt" for sample in samples]
+second_summaries = [f"results/stats/{sample}_after_recal.summary.txt" for sample in samples]
 rule all:
     input:
         recalibrated_files,
+        first_summaries,
+        second_summaries
 
 
 include: "rules/add_or_replace_rg.smk"
 include: "rules/mark_duplicates.smk"
 include: "rules/index_genome.smk"
 include: "rules/split_n_cigar_reads.smk"
 include: "rules/recalibration.smk"
+include: "rules/alignment_summary.smk"

Diff for: workflow/rules/alignment_summary.smk

@@ -0,0 +1,33 @@
+rule alignment_summary_before_recalibration:
+    input:
+        ref=reference,
+        bam="results/split/{sample}.bam",
+    output:
+        "results/stats/{sample}_before_recal.summary.txt",
+    log:
+        "logs/picard/alignment-summary/{sample}.log",
+    params:
+        extra="--VALIDATION_STRINGENCY LENIENT --METRIC_ACCUMULATION_LEVEL null --METRIC_ACCUMULATION_LEVEL SAMPLE",
+    resources:
+        mem_mb=1024,
+    wrapper:
+        "v3.12.1/bio/picard/collectalignmentsummarymetrics"
+
+
+rule alignment_summary_after_recalibration:
+    input:
+        ref=reference,
+        bam="results/recal/{sample}.bam",
+    output:
+        "results/stats/{sample}_after_recal.summary.txt",
+    log:
+        "logs/picard/alignment-summary/{sample}.log",
+    params:
+        extra="--VALIDATION_STRINGENCY LENIENT --METRIC_ACCUMULATION_LEVEL null --METRIC_ACCUMULATION_LEVEL SAMPLE",
+    resources:
+        mem_mb=1024,
+    wrapper:
+        "v3.12.1/bio/picard/collectalignmentsummarymetrics"
+
+
+