Read-Depth Profiling and Segmentation

[PatS]

Hi Tobias,

I have a few questions regarding the Delly tool that I would appreciate your insights on.

When generating read-depth profiles and performing CNV calling with Delly using these commands:

delly cnv -a -g hg38.fa -m hg38.map -c out.cov.gz -o out.bcf input.bam
Rscript R/rd.R out.cov.gz

which model of normalization is applied to the coverage data? Additionally, when plotting the output file out.cov.gz using R to generate normalized copy-number profiles, how is the read-depth information segmented? Does Delly use a Hidden Markov Model for this segmentation?

We recently obtained different plots from CNVkit, which showed worse segmentation, and we are trying to understand the differences.

Thank you for your answer!

Best regards,

Patricie

[Tobias Rausch]

Hi,

Delly uses fragment-based GC normalization as described in https://doi.org/10.1093/nar/gks001  and a mappability map for GRCh38 derived from dicey chop & map ( https://github.com/gear-genomics/dicey )

The segmentation method uses circular binary segmentation ( https://doi.org/10.1093/bioinformatics/btl646 ). The sdUNDO parameter ( https://github.com/dellytools/delly/blob/main/R/rd.R?plain=#L14 ) influences the granularity of the segmentation.

Best, Tobias

--
You received this message because you are subscribed to the Google Groups "delly-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to delly-users...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/delly-users/2a6c63bf-8e87-4b42-9e91-197c28fbf291n%40googlegroups.com.

[Tobias Rausch]

Hi,

It is possible to call CNVs on chrY to some extent (outside the heterochromatic and centromere regions).

I would recommend to use the CHM13 reference for this: https://s3-us-west-2.amazonaws.com/human-pangenomics/T2T/CHM13/assemblies/analysis_set/chm13v2.0_maskedY_rCRS.fa.gz

For short-reads, the matching delly mappability map is then: https://gear-genomics.embl.de/data/delly/T2T-CHM13v1.1.fa.r501.s2001.gz

Best, Tobias

On Tue, Oct 29, 2024 at 12:26 PM PatS <john.p...@gmail.com> wrote:

Hi Tobias,

thank you for your answer.
I've noticed that the rd.R script now includes the chrY chromosome (commented out).
Does this mean it is now possible to include chrY in read-depth profiling and segmentation plots, or is this still in progress? If so, which reference should be used for CNV calling?

Best,
Patricie

Dne čtvrtek 24. října 2024 v 11:06:50 UTC+2 uživatel Tobias Rausch napsal:

[tr]

Sorry, that was the wrong link for the CHM13 mappability map: https://gear-genomics.embl.de/data/delly/T2T-CHM13v1.1.fa.r101.s501.gz

[Sounkou Mahamane Toure]

Hi,
I wanted to add a question with regard to this. Given that XY samples are expected to have 1 copy of Y and X, does delly cnv calls cnvs on the Y, X chromosomes when call with ploidy 2 ? Is there a way to call autosome wide and X,Y chromosomes with delly at the same time ?

Thanks

[Sounkou Mahamane Toure]

Hi,

I have a question related to this. Given that XY samples have expected copies of 1 on X,Y chromosomes, is there a way to call at the same time CNVs on the autosome and X, Y  chromosome when ploidy is specified ?

Thank you