Yesterday’s monthly webcast featured our end-to-end solution for clinical labs. In case you missed it, you can watch the webcast here! There were a number of questions asked in the live event, so in today’s blog post I wanted to recap the Q&A session.
Can you go over the requirements for the CNV algorithm in more detail? If you are running CNV on a gene panel or exome, we recommend you have at least 100x coverage on average. We also recommend you have at least 30 reference samples; the algorithm does require a minimum of 10. Also, the algorithm does work under the assumption that these samples are from the same platform and library preparation. If you are going to be calling variance in non-autosomal regions, we do recommend you have gender match controls in order to perform those calls.
Does you CNV calling work on smaller CNVs? What is the relationship between coverage and CNV size? Yes, it does work on smaller CNVs, especially if you are calling on exons or gene panels. We can get down to a resolution of around 100 base pairs. Ultimately, it’s based on the size of your targets. The smallest CNV we can call is equal in size to the smallest target that you have in your set of targets when you compute coverage
How does the solution architecture work for an entirely automated secondary pipeline? The idea is that you can very easily integrate both Sentieon’s solution for calling variance with our CNV algorithm, in fact, in our next release, we are going to have a CNV command line runner so that you’ll be able easily to integrate within a batch script, the ability to call both variance using Sentieon as well as copy number variations and LOH events using our CNV caller. This can be completely automated from the command line using batch scripts and Linux.
When analyzing TRIO data can you call CNVs on mother, father, and proband and analyze them in VarSeq? And can I analyze quads as well? Yes, you can. When you have a VarSeq project you can call CNVs on any samples in that project. By default, it will run the algorithm on every sample in the project and you can even plot all of the CNVs for all the samples at once and even cross compare all of your samples.
Can you elaborate on the API a bit more. What technology stack is supported? Warehouse API is designed to be as versatile as possible. Essentially, you’ll send an http request to the warehouse server containing a query specified in JSON. It will return back an http request with the results of your query also stored in JSON. You can work with this API in any programming language that is capable of sending http request and is capable of parsing JSON; it can integrate with virtually any system in any programming language. If you need more fine grain control than the built-in API offers, you can also interact directly with the database which is a postgresql database. Normal SQL queries would be able to interact with that.
If you have any more questions or would like further clarification, please contact us at info@goldenhelix.com!