Significantly reduced DEL SUPPORT

weishwu commented 5 months ago

I ran Sniffles2 to call SVs from my amplicon ONT data. In one of the samples there is a strong ~34bp deletion that is carried by more than half of the reads:

As shown on IGV, this deletion starts from position 188 and ends at 221. However in the Sniffles2 output, this deletion was called as:

C392    181 Sniffles2.DEL.FS0   GTCATAGTATAACTTCGTATAATGTATGCTATACGAAGTTA   60  STDEV_POS   IMPRECISE;MOSAIC;SVTYPE=DEL;SVLEN=-40;END=221;SUPPORT=89

Sniffles2 shifted the start position to 181 (and the DEL length became 40, which may be caused by some reads that have a longer DEL at this locus). What is more concerning, the SUPPORT is 89, hugely lower than the truth as shown on IGV.

I filtered out the alignments that had mapq<20 so it can't be because of that.

Below are all the SVs called by Sniffles2. No neighboring DEL can explain the loss of SUPPORT:

Contig  POS END TYPE    SVLEN   SUPPORT
C392    44  92  DEL -48 11
C392    48  48  INS 12  47
C392    48  48  INS 30  16
C392    48  48  INS 64  3
C392    48  58  DEL -10 3667
C392    50  70  DEL -20 108
C392    118 118 INS 46  6
C392    119 119 INS 81  1
C392    153 245 DEL -92 5
C392    182 222 DEL -40 89
C392    195 204 DEL -9  158
C392    375 375 INS 117 2
C392    458 458 INS 9   45
C392    460 460 INS 27  12
C392    480 480 INS 53  8
C392    732 800 DEL -68 4
C392    740 740 INS 64  1
C392    767 800 DEL -33 48138

My command-line:

singularity run -B /nfs/ sniffles_v2.3.3.sif sniffles \
       --sample-id test \
       --reference C392.fa \
       --input C392_891_span.sorted.aligned.clean.bam \
       --vcf test.vcf \
       -t 2 \
       --minsvlen 10 \
       --mapq 20 \
       --min-alignment-length 100 \
       --no-qc \
       --output-rnames \
       --minsupport 2

Any ideas? Thanks.

weishwu commented 5 months ago

Hi @hermannromanek , any thoughts/suggestions on this?

fritzsedlazeck commented 5 months ago

Thanks @weishwu
is it possuble to share the bam file with us ? Or maybe make a screenshot of the IGV? My email: fritz.sedlazeck@gmail.com

I suspect that there might be something else going on as Sniffles ignores some reads due to quality. Have you tried to just run with default parameters?

Thanks Fritz

hermannromanek commented 5 months ago

Hi @weishwu

Sorry for the late reply - as Fritz suggested, sharing a (regional) bam file could help us identifying whats going on here. With the next version of sniffles we included the fix for an issue with with very clean cut deletions, so this may already be fixed as well, but we'd have to double check it.

Thanks, Hermann

weishwu commented 5 months ago

Hi @fritzsedlazeck I've shared a globus link with Fritz (fritz@access-ci.org). The folder contains the BAM, index, and the VCF. Let me know if you want me to use a different way to share the data. https://app.globus.org/file-manager?origin_id=af373541-899f-4eed-bae9-8ff738475f2d&origin_path=%2F @hermannromanek Sorry I can't find your account on globus. Thanks a lot for your help!

hermannromanek commented 4 months ago

Hi @weishwu

I registered through my Github account on globus - can you please also share the directory with me?

Thanks, Hermann

fritzsedlazeck / Sniffles

Significantly reduced DEL SUPPORT #484