search

cgab-ncc / FIREVAT

FInding REliable Variants without ArTifacts
MIT License
21 stars 8 forks source link

Error in fisher.test(test.mat) when using Strelka format #6

Open ozlemmuslu opened 4 years ago

ozlemmuslu commented 4 years ago

Hi,

Thanks for publishing this tool.

I tried running it using VCFs from different variant callers, however, I get stuck with this error:

` [1] "INFO [2020-05-18 20:18:50] Step 03. Additional analysis."

[1] "INFO [2020-05-18 20:18:50] Step 03-1. Perform strand bias analysis [firevat_strand_bias::PerformStrandBiasAnalysis]"

Error in fisher.test(test.mat) : all entries of 'x' must be nonnegative and finite `

I've read your documentation but couldn't find instructions on how to fix this.

khb7840 commented 4 years ago

Hi @ozlemmuslu!

"PerformStrandBiasAnalysis" requires strand-specific read count for each variant allele; Reference-Forward, Reference-Reverse, Alternate-Forward and Alternate-Reverse.

As the default output format of "Strelka" does not provide any information on strand-specific read counts (https://sites.google.com/site/strelkasomaticvariantcaller/home/somatic-variant-output), strand bias analysis is not available with the default FIREVAT config file for Strelka-called vcfs.

Setting "perform.strand.bias.analysis = FALSE" when running "RunFIREVAT" may help.