In this vein, we have tackled problems in CRISPR gene editing problems in statistical genetics such as effective and efficient handling of unknown confounding factors in eQTL association studies, genome-wide association studies, and analysis of methylation data immunoinformatics such as HLA imputation and refinement, epitope prediction problems in proteomics such as alignment of vector time series resulting from liquid-chromatography-mass-spectrometry systems. We also approach many problems from the perspective of applied statistics and machine learning, making use of latent variable models and efficient operations on them to perform inference and learning. We encompass several approaches to computational biology: we try to frame the biological question under consideration in terms of more standard problems in computer science, like clustering, Steiner trees, flow problems, etc., and then use approximation algorithms motivated by statistical physics to solve these problems. One of our most successful approaches in this realm involves variants of belief- and survey propagation algorithms, but in the course of adapting our problem to this setting, we often need to derive alternative representations of the original computer science problem which might be useful when applying other algorithms as well.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |