Hello all and many thanks for all the answers I have read here in the past.

I need some feedback from the community. How do you handle SNP calls in empty wells?

I receive illumina sequencing files with 96 wells, our lab leaves one well empty. When I process the files with a SNP calling pipeline there are reads in the empty well. usually about 2 standard deviations below the mean of read counts for the other 95 well. But there are reads there and they do process to become SNP calls.

I think that these 'reads' simply represents noise in the data, and when noise gets compared to a reference genome, I expect that 75% of random would be assigned the status of SNP.

I usually just ignore these SNPS and drop them before prediction, how would you approach this situation? Should I be concerned at all?

Thanks!