Custom made Panel of Normals required?
For what I understand, the GATK's best practices recommends the usage of a Panel of Normals (PON) for an accurate somatic mutation detection. By having a large enough pool of "normal" samples, it is possible to detect technical biases created by the sequencing machine, the protocol, the kit, etc. Please correct me if I'm wrong.
With this said, I'd like to analyse WGS data coming from BGI, using DNBseq platform instead of the industry standard Illumina. If I do not have a set of "normal" samples sequenced in the same way as my WGS data, is it accurate to use a generic PON (even if it's data was generated from an Illumina machine)? Or is it more accurate to not even use any PON at all?
Hi Pedro Miguel Raposo, we would recommend creating a PON using what little BGI data you have. The data from the BGI sequencer is likely to be significantly different from Illumina data in terms of error modes, so it is unlikely to work well if you use a generic PON. It could end up filtering out real sites and cause a slight loss in sensitivity.
Please sign in to leave a comment.