search

cortes-ciriano-lab / SComatic

A tool for detecting somatic variants in single cell data
Other
145 stars 20 forks source link

Application to Smart-seq2 #13

Closed LiQian-XC closed 11 months ago

LiQian-XC commented 1 year ago

Hi Francesc,

Thanks for this useful tool. When applying SComatic to single-cell data from Smart-seq2, are the beta-binomial parameters needed to be adjusted/re-inferred using a panel of unrelated samples from Smart-seq2? If yes, can you recommend which samples to use for estimating these parameters, especially when I want to call somatic mutations from normal samples/tissues? Similarly, is PON also needed to be customised?

Thank you in advance!

Best, Qian

yuhsinhsieh-josch commented 1 year ago

Hi Francesc,

I have similar question, but would like to use SComatic for scATAC from 10X genomic.

Thank you!

Best, Yu-Hisn

Francesc-Muyas commented 1 year ago

Dear users,

Thanks for bringing up this interesting point about the Smart-seq2 data. Although we have not tested SComatic with such data, I do not see the reason (as far as they have the CB information in the bam) for not being suitable for our algorithm.

As you have mentioned, it would be better to re-estimate the Beta-Binomial parameters with tumour-free samples, which are expected to have low numbers of true somatic mutations. If you cannot find enough samples (at least 5-10) for this estimation, I would suggest going ahead with the default Beta-Bin parameters but increasing the _--min_accells parameter in Step 4.1 to 3 ( _--min_accells 3). In addition, the ideal situation would be to re-construct your own PoNs with multiple samples. Nevertheless, we know that this is not always possible. If so, you could use one of the PoN potions described in Step 4.2.

For the scATAC-seq data, you can be quite confident using the default Beta-Binomial parameters in SComatic and the scATAC-seq PoN in this repository. However, you can also use other PoN alternatives (e.g. provided by GATK).

Thanks, Fran