SMBE 2014 Day 2

Day 2 in Puerto Rico.

Lots of exciting talks today!

Some highlights:

– Fernando Racimo from UC Berkeley asking “how far in the past can we detect positive selection on the human genome?” by comparing fixed differences between humans and Neanderthals. Started out by demonstrating that there is significantly lowered scaled diversity around substitutions affecting splice sites in humans (relative to synonymous fixations), but background selection masks any signal from most other substitutions. He expanded CLR test for natural selection to include Moran model of sequence evolution and other additions that allows it to be used with larger periods of genetic drift separating populations (closely related species possibly!) to get at detecting ancient selection in a different way.

 

– Patricia Wittkopp (Michigan) presented an interesting approach to compare how mutations introduced by EMS in yeast affect gene expression differently in cis and trans. She looked in particular at the TDH3 promoter: turns out most new mutations have small effects on gene expression which are likely nearly neutral. However, gene expression noise should increase drastically with new mutations in the promoter. Since the level of noise expected without selection is not detected in natural populations, noise seems to be selected against.

 

– Katherine Pollard (UCSF) presented great work that is focused on using the myriad of available data to improve our accuracy of identifying functional regulatory elements and interactions in humans. “EnhancerFinder” is a support vector machine approach to combine many different heterogeneous data-types, which was used to call 80,000 enhancers. TargetFinder can discriminate true/false chromatin conformation interactions based on several features, the best of which are TF binding events and histones marks in between the called enhancer and promoter. Lastly, motifDiverge sounds like a promising model for testing for differences in the motif composition between any two sequences.

 

– Gavin

Leave a Reply