This year the MAGES symposium has a new name, the Symposium on Advances in Genomics, Epidemiology and Statistics or SAGES! The NIH has also been added to the sponsor list and we’re excited to have the support for this informative symposium! Additionally, we were treated to the addition of a new poster session to accompany the fantastic speakers.
The trend for this year’s speakers was data integration; pulling various data types together into methods and interactions to try to understand the underlying biology of disease and discover previously unidentified associations. Here is a snapshot of the methods and speakers that caught my attention.
Dr. Richard Bonneau started off the day discussing the SPIEC-EASI (Sparse Inverse Covariance Estimation for Ecological Association Inference, Kurtz et al. 2015) method for mapping microbiome associations along with integrating the method with immunological data such as from ChIP-Seq.
Dr. Mingyao Li from UPenn presented a method for expression quantification of alternative splicing variants or isoforms in RNA-Seq data, a method dubbed PennSeq (Hu et al. 2014). This method doesn’t assume a data distribution, but let’s the data “speak for itself!” Most methods assume read start position is uniformly distributed, but this method can handle non-uniform read distributions. She also presented MetaDiff, a new approach to differential isoform expression analysis that uses random-effects meta-regression; the method paper has been accepted by BMC Bioinformatics and will be published later this year.
Dr. Stephen Montgomery presented a family-based way of looking at the impact of rare noncoding variants. He presented a large family with eleven children to measure the impact of rare variants in the family and detect eQTL’s through linkage (Li et al. 2014). Through the rare variants predicting family effects was possible. This was also demonstrated using the Sardinia population dataset, which includes family information, for the following diseases: MS, Type I Diabetes and Malaria. This example also showed that conservation scores proved more useful then prediction algorithms for identifying variants in these diseased families.
Dr. Nancy Cox presented a method the draws from genetic data, genetically derived expression (GReX) and trait information, called PrediXcan. The method uses the theory that a substantial fraction of the genome variation affecting risk of common disease is regulatory and aggregates variants into SNP-based predictors of transcript levels. PrediXcan is a gene-based test that associates genes with causal phenotypes (Lonsdale et al. 2013 , Gamazon et al. 2015).
This year SAGES had the most diverse group of researchers to date. Paired with the added support of the NIH and leaders in the realm of statistical genetics presenting such forward thinking topics, this symposium will continue to grow! I can’t wait to see what SAGES 2016 will bring.