Convert AB Sanger traces to FASTQ and align

ab2bam

#!/bin/bash

# Unzip files
for file in *.zip; do
        unzip "$file"
done

# Extract fastqs
for file in *.ab1; do
        seqret -sformat abi -osformat fastq  -auto -stdout -sequence "$file" > "$(basename "$file" .ab1).fastq"
done

# Map reads to reference
#bwa_index=/resources/hg19/bwa/0.7.5/ucsc.hg19.fasta
bwa_index=/resources/b37/bwa/0.7.5/human_g1k_v37.fasta
for file in *.fastq; do
        id=$(echo "$file" | cut -d . -f1)
        sample=$(echo "$file" | cut -d _ -f1)
        bwa mem -t 4 -R "@RG\tID:${id}\tSM:${sample}\tLB:${sample}\tPL:ILLUMINA\tCN:OMICRON" ${bwa_index} "$file" | samtools view -Sbh - -o "${id}.single.bam"
done

# Merge bams for same sample
for sample in $(ls -1 *.single.bam | cut -d _ -f1 | sort -u); do
        samtools merge ${sample}.bam ${sample}*.single.bam
        samtools index ${sample}.bam
done

# Cleanup
rm -rf *.seq *.fastq *.single.bam