Issue with the detection of deletions that I see are visible

[tanyasarkjain]

Describe the bug
Hi I am running:

CRISPRessoWGS -b "reads.bam -f vegfa.tsv -r hg38.fa --name "output" --exclude_bp_from_right 0 --exclude_bp_from_left 0 --quantification_window_size 10 --min_frequency_alleles_around_cut_to_consider 0.001

Expected behavior
Visually looking at the produced ANALYZED_REGIONS folder, I can see there is a deletion that goes into the target site in my tsv, (the target site is 20bp) however, the deletions are not being counted when I look at the indels, perhaps because it goes beyond the target region. However, even after extending --quantification_window_size 10, and the target location to span the entirety of the deletion, it is still not being identified. Is there another parameter I should be running?

[kclem]

Hi @tanyasarkjain

Are you using short or long-read sequencing?

How long is your deletion? If the deletion is more than 60% of the length of your region in -f file, it won't align and will be discarded. You can change this behavior with the parameter --default_min_aln_score.

CRISPRessoWGS will only include reads that completely span the region in your -f file. It does this by chopping all reads down to start and end at the regions in your -f file. Unfortunately, if the start or end is deleted, the read may not be included. Could this be the case? Is the deletion completely in your -f region or does it overlap?

Alternatively, you can try this branch: https://github.com/pinellolab/CRISPResso2/tree/wgs-whole-region-deleted which allows for analysis when the start or end (or the whole region) is deleted. We're still testing this branch, so if you have any feedback, that would be great.

Otherwise, these are the steps I'd take to debug:

1. Identify the read id of the read with the deletion using IGV or another viewer
2. Check to see if the read made it into the ANALYZED_REGIONS/*.bam and fastq.gz files
3. Run with --fastq_output to see the alignment and the alignment score for the read

Feel free to reach out if you need additional help or want to walk through this together - [email protected].

[tanyasarkjain]

Hi @kclem , thank you for the response.

Here is a segment of the reads -- there are more below that are not visible in this screenshot.

They are short reads and the reads do span the whole region - however the deletion extends past the target region. I am using BAMs not fastqs also as my input. Thank you! The red indicates the target region, and the deletions seen in the ifrst few reads are not being counted). All these reads where in the ANALYZED_REGIONS folder for the associated region so they seemed to make it in the ANALYZED_REGIONS/*.bam, but when I look and the "deletions" stats in the SUMMARY txt file, I only see a deletion of 1

[kclem]

Hi @tanyasarkjain,

Yes, it looks like the reads are being discarded because they deletions are not contained within the region.

Can you expand your window to include the bases surrounding the deletion?

Are you able to try this branch which allows these types of reads? https://github.com/pinellolab/CRISPResso2/tree/wgs-whole-region-deleted