High Number of Deletion Calls and Correlation Discrepancy in ExomeDepth Analysis

[valeandri]

Hello @vplagnol,

I hope this message finds you well. I recently performed an ExomeDepth analysis on a targeted sample, using a reference set comprising 10 male samples. Surprisingly, the analysis yielded an unusually high number of deletion calls (2016 deletions and 131 duplications). Despite this, the correlation with the reference.choice, consisting of 4 out of the 10 samples, exhibited a good value of around 0.99.

Upon closer inspection, I observed that the test sample's coverage was approximately 10-20% lower than that of the samples in the baseline. Most of the deletion calls turned out to be false positives. For your reference, I have attached screenshots from IGV comparing the test sample with those in the reference.

My primary concern is understanding why the sample exhibited such a high correlation despite its significantly lower coverage. Typically, in cases of lower coverage, one would expect a lower correlation, enabling the easy identification of problems.
Is there a specific reason for this unexpected behavior?
Additionally, could this pose a higher risk of false negatives among duplications?

I have provided basic information, and I am more than willing to share additional files and logs if required.

Version used: 1.1.16

Looking forward to your insights and guidance on this matter.

Best,
Valentina

examples_nonames.pdf