BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Computational methods for analysis of single-cell RNA-seq and canc
 er mutability - Martin Hemberg (Sanger Institute)
DTSTART:20170323T120000Z
DTEND:20170323T130000Z
UID:TALK71735@talks.cam.ac.uk
CONTACT:William Grant
DESCRIPTION:In this talk I will be presenting some of the work in my group
  focusing on quantitative models of gene regulation. The first part will c
 over computational analysis of single-cell RNA-seq data\, while the second
  part will cover the role of non-canonical secondary structures of DNA on 
 mutability in cancer patients.\n\nCompared to bulk RNA-seq\, single-cell R
 NA-seq presents many opportunities for addressing biological questions tha
 t were previously inaccessible. To realize this potential\, however\, nove
 l computational methods are required. We have developed a new method for u
 nsupervised feature selection. Without any prior assumptions of cell-types
 \, the method is able to identify a set of informative genes. We demonstra
 te that these genes correspond to differentially expressed genes and that 
 they are more informative than highly variable genes for clustering and ba
 tch correction. Moreover\, we use feature selection for scmap\, a fast and
  accurate method for comparing two different samples. \n\nAlthough DNA for
  the most part is in the canonical B-DNA configuration\, there are more th
 an 20 other configurations known. Here\, we demonstrate that the non-canon
 ical secondary structures are predictive of mutability in cancer. The incr
 eased mutability of sites overlapping non B-DNA motifs suggests that one m
 ust take this aspect into consideration when characterizing driver mutatio
 ns.\n\nIn addition to presenting recent results from my group's research\,
  I will also discuss some of my thoughts on how people from a physics back
 ground can make contributions to molecular biology and genomics.
LOCATION:TCM Seminar room\, 530 Mott building
END:VEVENT
END:VCALENDAR