Yesterday’s webcast, Genomic Prediction Methods in SVS, gave attendees a chance to see how the principles of genomic prediction are applied within SVS, predicting phenotypes for both plant and animal species. You can find a recording of the webcast on our site here should you be interested in checking it out or sharing with a colleague!
The webcast garnered a lot of interest and some great questions. We thought we’d share these with the community in case you were wondering the same thing…
Does the program allow for fitting fixed effects in GBLUP?
Yes, you can account for additional covariates into your model and we just recently added methods necessary to account for environmental interactions as well.
Do Bayesian methods consider relatedness?
Yes, we didn’t get look at prediction output for Bayes C or Bayes c pi, but you will generate a relationship matrix just like with GBLUP
Does SVS use all markers to predict the breeding value for oligocentric traits?
It depends on the model you use. With GBLUP, it will use all the markers you give it. Unfortunately, this webcast doesn’t cover it. But, you can filter to high-quality markers on-call rate for example or maybe allele frequency, and include those filtered markers in GBLUP. With Bayesian methods, it will just identify the most influential markers to build the model.
Is this part of standard SVS or an add-on capability?
This is part of our standard package. All the prediction methods, as well as our k-fold cross validation, are included in the SVS power seat package. For larger sample sets we recommend a server license. If this is something you’re interested in, email info@goldenhelix.com. We’d be more than happy to discuss details with you.
Are you planning additional prediction methods?
Currently, we feel there is a sufficient number of methods that come with our product. We are aware that there are a plethora of methods in the Bayes family. In some cases, it really just comes down to preference. For very large clients we would consider additional development work as part of a larger license arrangement.