Merge pull request #52 from Juke34/he

Adding hyper editing processing

Author: Juke34

bin/restore_sequences.py

@@ -0,0 +1,47 @@
+#!/usr/bin/env python3
+"""
+Restore original sequences in BAM files from original unmapped BAM.
+This tool replaces modified sequences with their original counterparts.
+"""
+
+import sys
+import argparse
+import pysam
+
+
+def fix_bam(bam_aligned, bam_unmapped, bam_out):
+	"""Restore original sequences using indexed lookup."""
+	# Create index for fast read name lookup
+	unmapped = pysam.AlignmentFile(bam_unmapped, "rb")
+	unmapped_index = pysam.IndexedReads(unmapped)
+	unmapped_index.build()
+	
+	with pysam.AlignmentFile(bam_aligned, "rb") as bamfile:
+		with pysam.Samfile(bam_out, 'wb', template=bamfile) as out:
+			for read in bamfile:
+				try:
+					# Fetch original read by name (returns iterator)
+					for orig_read in unmapped_index.find(read.query_name):
+						read.query_sequence = orig_read.query_sequence
+						read.query_qualities = orig_read.query_qualities
+						read.set_tag("MD", None)
+						break  # Take first match
+					out.write(read)
+				except KeyError:
+					pass  # Read not found in unmapped BAM
+	
+	unmapped.close()
+
+
+def main():
+	parser = argparse.ArgumentParser(description=__doc__)
+	parser.add_argument("-b", "--bam", required=True, help="aligned BAM file")
+	parser.add_argument("-u", "--unmapped", required=True, help="original unmapped BAM file")
+	parser.add_argument("-o", "--output", required=True, help="output BAM file")
+	args = parser.parse_args()
+
+	fix_bam(args.bam, args.unmapped, args.output)
+
+
+if __name__ == "__main__":
+	sys.exit(main())

build_containers.sh

@@ -115,7 +115,7 @@ if [ "$build_docker" = true ]; then
     # List of image names
     image_list=( )
 
-    for dir in docker/*; do 
+    for dir in containers/docker/*; do 
         cd "${dir}"
         imgname=$(echo $dir | rev | cut -d/ -f1 | rev)
         image_list+=(${imgname})
@@ -161,7 +161,7 @@ if [ "$build_singularity" = true ]; then
     sif_list=( )
 
     # Loop through all .def files in singularity/ subdirectories
-    for def_file in singularity/*/*.def; do
+    for def_file in containers/singularity/*/*.def; do
         # Extract image name from directory
         imgname=$(basename "$(dirname "$def_file")")
         sif_output="sif_images/${imgname}.sif"

config/softwares.config

@@ -46,7 +46,7 @@ process {
         container = 'quay.io/biocontainers/star:2.7.10b--h6b7c446_1'
     }
     withLabel: 'samtools' {
-        container = 'quay.io/biocontainers/samtools:1.3.1--h0cf4675_11'
+        container = 'quay.io/biocontainers/samtools:1.23--h96c455f_0'
     }
     withLabel: "sapin" {
         container = singularity.enabled ? "${params.sifPath}/sapin.sif" : "sapin"

docker/jacusa2/Dockerfile containers/docker/jacusa2/Dockerfiledocker/jacusa2/Dockerfile renamed to containers/docker/jacusa2/Dockerfile

@@ -51,6 +51,7 @@ dependencies:
   - typing_extensions=4.13.2=pyh29332c3_0
   - tzdata=2025b=h78e105d_0
   - wheel=0.45.1=pyhd8ed1ab_1
+  - pysam=0.23.3=py312h47d5410_1
   - pip:
       - flameprof==0.4
 prefix: /home/earx/miniforge3/envs/rain-stats-scripts