Research
Follow the strand.
Then pull the thread.
Narrowing the gap between
lifespan and healthspan.
Featured
-
Methylation Changes in the Peripheral Blood of Filipinos with Type 2 Diabetes Suggest Spurious Transcription Initiation at TXNIP
In the blood of Filipinos with type 2 diabetes, 177 methylation sites shift. Three sit in TXNIP, where the change looks like transcription starting in the wrong place.
-
Correlation-Adjusted Simultaneous Testing for Ultra High-Dimensional Grouped Data
When methylation probes are tested in correlated groups, the usual group-wise correction lets false discoveries through. CAST keeps the false discovery rate in check, and found probes in bladder cancer data an earlier study missed.
-
Empirical Null Estimation using Zero-inflated Discrete Mixture Distributions and its Application to Protein Domain Data
Mutation counts on protein domains carry more zeros than the textbook null allows. A zero-inflated generalized Poisson null picks out the counts that matter at a fixed false discovery rate.
-
Bayesian Local False Discovery Rate for Sparse Count Data with Application to the Discovery of Hotspots in Protein Domains
With sparse mutation counts, standard multiple testing loses control of its error rate. A Bayesian local false discovery rate finds protein domain hotspots across whole gene families.
-
Oncodomains: A Protein Domain-Centric Framework for Analyzing Rare Variants in Tumor Samples
A rare tumor mutation means little gene by gene. Compared across the shared domains of a gene family, clusters of them point to a role in cancer.
-
Bayesian Variable Selection using Knockoffs with Applications to Genomics
Knockoffs, the Bayesian lasso and the local false discovery rate in one procedure, to find which genetic markers drive HIV drug resistance while keeping the false ones out.