Feedback on mtDNA variant calling analyses

[caleblareau]

Hi @danieljrichard, thanks for sharing this. Overall, I think the experiment looks good, and I can vouch for the workflow and clone calls. A few things to think about:

• I'm curious what the CellRanger QC report looks like for these? What percent of non-nuclear reads did you get for this cell type? Since you're dealing with low heteroplasmy mutations, more mtDNA depth is probably going to extra critical for your samples.
• Clone 3 looks suspect; there are two mutations that are transversions (T>As) that are defining the clone but rarely co-occur; I wouldn't make too many conclusions on this one. This could be low-frequency sequencing error or alternatively a bit of a limitation in the clone caller.
• Looking at the strand correlation/ VMR plot, you could maybe go down as low as 0.55 where there is a cleaner separation in signal and capture a few more mutations. I would just be cautious with these if they aren't transitions (C>T/A>G/G>A/T>C)

Thanks for sharing!

[danieljrichard]

Hi @caleblareau

Thank you so much! I don't want to take up too much of your time, but to address your comments briefly:

1. Around 15% of reads were non-nuclear. Since we got way fewer cells than expected from the 10X, we sequenced each cell pretty deeply (>250K reads per cell)
2. That's a good point. Along with (3) it sounds like I should be more wary of transversions, especially at the lower strand correlation values.

Thanks again!