Off-target % for whole-exome sequencing panel

Question

[bompipi95]

Hi all,

My samples have been sequenced with the Twist Exome 2.0 sequencing panel, and I want to assess how efficient the sequencing has been. By efficient, I mean examining metrics such as the uniformity (fold-80 score), coverage, on/off-target % etc. as documented in their webpage. I checked the % off-target as follows:

1. Count the # of reads in the bam file (samtools view -c "$dir/$base.bam)
2. Count the # of reads that intersect with the bed region (bedtools intersect -a "$dir/$base.bam" -b $bed -u | samtools view -c -)
3. Calculate the % off-target rate as (bc -l <<< "(1 - ($bed_reads / $total_reads)) * 100")

I have two questions:

1. Is there a cleaner/simpler way to calculate the off-target rate? I know I can calculate the on-target rate with bedtools coverage, but I am not sure of how to calculate the off-target rate with a single command line.
2. I obtained an off-target rate of approximately 20-25% for my samples. Is this within the "good" off-target % range?

[Pierre Lindenbaum]

better/faster ?:

samtools view -F 3840  -c "$dir/$base.bam
samtools view -M -L capture.bed -F 3840  -c "$dir/$base.bam

Answer 1

[a.alnawfal.1992]

for the first question, please see Picard tools like CollectHsMetrics:

CollectTargetedPcrMetrics

Tool doc

For second question we need more information to make a clear judgment on the sample

Answer 2

[trausch]

With alfred it would be a single command:

alfred qc -b hg38.twist.bed -r hg38.fa -j qc.json.gz -o qc.tsv.gz wes.bam

Depending on the WES kit, off-target reads are often close to target regions. That's why, alfred calculates directly on target (+/-0bp) and then gradually extends target regions. The JSON while can be uploaded at https://www.gear-genomics.com/alfred/ for interactive browsing or on the command-line:

zgrep "^OT" qc.tsv.gz