Jeffrey M Kidd

We hope the CNV data we released we will be useful to others, and look forward to applications that use the many other improved genome references now being constructed.

There are many other interesting examples of gene parlog variation in these samples. We think QuicK-mer2 will enable the routine assessment of such differences in the huge number of sequenced genomes now available.

Of course, this is only a copy-number estimate. The structure, sequence, and potential functional impact of this rare variation remains to be demonstrated.

Coming back to where this thread started: when you look across thousands of samples it is not surprising that you find rare variants that were previously missed.

Here in 2020 it is remarkable how useful the 1000 Genomes Project remains. The freely accessible data and samples continue to be the standard for uncountable analyses of human genetic diversity.

With this increased speed, we were able to create paralog-specific maps from the 2,500 samples from the 1000 Genomes Project. We made use of the newly sequenced 30X coverage samples made available by the NY Genome Center @nygenome @1000genomes @genome_gov

It relied upon external tools for the core processing, which required the use of large intermediate files that really slowed things down. It was also somewhat unwieldly, and required complicated steps to identify appropriate sets of k-mers to analyze from each reference genome

Like many in genomics over the past few years, we moved away from standard mapping to consider approaches based on ideas around k-mer counting. Feichen implemented this approach in a tool called QuicK-mer.

Given the huge explosion in short-read genomes, we wanted an approach that was fast and easy to use, with a focus on a paralog-specific view of copy number variation.

The field of CNV analysis is very large, and there are many other methods for short-read analysis with specific strengths. Several of the best integrate multiple types of read signatures to find diverse types of structural variants.

One major limitation is that this approach is blind to variation among individual paralogs. Due to the purposeful multi-mapping, the reads from each copy are lumped together.