Cluster assignment probability filtering

[rpensch]

Hi!

In the documentation you suggest that the cluster assignment probability can be used as a confidence score to filter out low probability mutations. I was wondering if you can recommend a threshold to use for filtering?

Edit: I actually came up with a second, unrelated question. In #14 you mentioned that when using multiple, related samples one has to provide read counts for positions that were not called as a variant in other samples. I have done some research and not found an efficient way to get these counts for reference and alternative allele. I was planning to approximate by getting the total number of reads at those positions which is done easily and then using ref_counts = total number of reads and alt_counts = 0, which assumes that if the mutation has not been called, it does not exist. Can I work with this assumption without affecting the results too strongly?

Thanks!

[sb43]

https://github.com/cancerit/vafCorrect
This tool will get ref and alt count at matching locations in related samples.