get_Aneuploidy_score for exome

[joonan30]

Hi Shixiang,

Thanks for the great tool. I've been using get_Aneuploidy_score for CCLE segmentation file and it seems working well as I compared with the original paper's result (Sharir et al. 2021).

Then, I tried this to my exome data and it seems pretty low AS score. Do you think this AS score method is not suited to exome segmentation?

Many thanks in advance

[ShixiangWang]

Hi, the key computation logical is:

• Take the segment-size weighted median total copy number of each chromosome arm (arm_wMedian)
• Assign the arm_wMedian a loss/neutral/gain (-1/0/1) status based on relation to the cell lines ploidy
• Using a 0.5 threshold (from the default round() function), determine whether the arm_wMedian is >, =, or < than the ploidy

from: https://github.com/quevedor2/aneuploidy_score/#details-of-aneuploidy-scorecaa

As exome only captures part of genome, it would expect that the AScore was lower that the value from SNParray/WGS (the length of copy number variation is not accurate when using exome). I don't validate this, if you have both the data, maybe you can take a comparison.

Another way is changing the method to calling the copy number variants, you may take a try and see how it works. What't tool are you using?

[joonan30]

Thanks for the comment. I will try this and let you know.

I am using CNVkit and facets (cnv_facets). For your method, they both give me similar result (low score ~ 5-8). For other methods (including your CNV signature analysis), both tools seem working and similar results to each other.

I don't have WGS for the same sample though.. My best guess for now is to check whether I have similar the WGD rate with previous result. I've used infer_wgd.py (https://github.com/vanallenlab/facets/tree/master/infer_wgd) and found pretty similar rate of WGD in my samples, compared to previous study (https://www.nature.com/articles/s41586-020-03133-3).

[ShixiangWang]

@joonan30 I think 5~8 is not a very low value, it means 5-8 chromosome arms are significantly altered. Perhaps you can compare the mean level with the same cancer types from previous study or TCGA (TCGA have processed copy number data from ASCAT).

[joonan30]

Thanks! I will check it out.