Gabe Rudy gave a fantastic presentation yesterday on the latest additions to VS-CNV annotations. If you weren’t able to join us for the live event, you can access the recording and webcast slides here: Comprehensive Clinical Workflows for Copy Number Variants in VarSeq. Additionally, there were many great questions asked that we wanted to share with the community.
Question: Can I import my existing cytogenic-based CNV annotations into the assessment catalog?
Answer: This is a great question, and I want to tie it into another question asked today, “Are you able to import custom annotations such as highly conserved exons or other historical annotations from our cytogenetic lab?” The answer is yes, all of these scenarios are supported. You can use our annotation Convert Wizard to create a more static annotation source, which can import BED files as well as delimited text files. Or, if you want a live-updating catalog that changes over time, use the Assessment Catalog feature. Our Assessment Catalog has a bulk import feature. I demonstrated in the webcast how this could bring in data from your tables in VarSeq, but you can also bring things in from text files. Any file with genomic coordinates can be used to bootstrap your catalog. And, you can set customize the schema to contain whichever fields you want to capture. These catalogs are both accessible in the interactive view in VarSeq as well as annotations and plottable sources to aid your interpretation.
Question: For the CNV calling, I heard there are two modules you used for small and large CNV respectively. When we use it, will they be utilized automatically or manually?
Answer: The whole process is very streamlined and automatic, so internally we use a lot of different heuristics and input data types. Some of those are going to have more heavyweight when you’re looking at a single exon event. And then a very different strategy starts to kick in on larger cytogenic sized events and chromosome aneuploidy events. For example, the single-exon events will be called by our probabilistic model using multiple heuristics, and larger events must pass a segmentation algorithm threshold but also considers other overlapping smaller events, loss of heterozygosity and allelic imbalance as well. This is all handled automatically providing a list of all called events in one table ready for annotation and interpretation.
Question: Does this new capability require an exclusive license?
Answer: This will require that you have a license to use our VS-CNV 2.0 caller product. You can get it as a bundle with VarSeq. All you’d have to do is reach out to your account representative or bring up your interest in evaluating the CNV capabilities when you are evaluating VarSeq in general.
Question: Why would a particular gene be prone to false positives?
Answer: These could be things like pseudogenes that themselves have copies or very similar homologous copies inside the genome. With the nature of short-read NGS sequencing is the genomic mapping requires some level of uniqueness of that short sequence. And around the boundaries where that uniqueness comes and goes you start to see highly variable coverage regions which can be prone to false positive events. But these genes can still be analyzed with special considerations. We found that PMS2, are very commonly annotated genes in certain gene panels, has a known pseudogene that makes the read coverage more variable. But we have a clinical customer who was able to validate it as part of their gene panel with high confidence calls by just ignoring some of our QC flags which report “Hey, we’re kinda within some noise here!” while still providing the raw CNVs needed. So you can still pick out the true positives versus the false positives – they just need a little more discerning effort, and your annotations may be used to flag those regions.
Question: Can you import CNV calls for samples from other external programs into VarSeq? Maybe from whole genome sequencing or from a microarray platform for the same sample?
Answer: This is something we are interested in supporting, but it will require a lot of test cases as there is no standard way to represent CNV calls. For example, we can import variants based on the specification of a VCF file. There is no consistency between say how the Affymetrix ChAS studio exports CNVs and Illumina GenomeStudio. I would be interested in working with customers who want to support that use case, especially if you’re able to share some example files. If we have interest in this, we will look into supporting bringing in these external CNV calls into a table similar to the one we looked at today, where you can use all the annotation and filtering capabilities we demonstrated as well as integrate the interpretation with that of your standard NGS variant analysis.