BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:BSU Seminar: “Unsupervised substructure discovery in mass spectr
 ometry metabolomics data” - Dr Simon Rogers\, University of Glasgow
DTSTART:20181108T140000Z
DTEND:20181108T150000Z
UID:TALK114448@talks.cam.ac.uk
CONTACT:Alison Quenault
DESCRIPTION:Identification of molecules observed in data derived from mass
  spectrometry experiments remains is very difficult\, and hinders the wide
 r application of high throughput untargeted metabolomics. The standard ana
 lysis pipelines compare observed spectra with databases of known molecules
 \, but these databases have very low coverage resulting in the majority of
  the measured molecules being unidentified. Here\, I will present an alter
 native approach for exploring datasets from untargeted metabolomics experi
 ments that uses a technique from text mining (Latent Dirichlet Allocation\
 ; LDA) to decompose the observed spectra across a set of  shared component
 s. We show that these often represent molecular substructures. In other wo
 rds\, we are able to break unknown molecules down into building blocks\, m
 any of which can be identified. I will show results on a number of standar
 ds and real datasets\, as well as describe some future directions of this 
 work.
LOCATION:Large Seminar Room\, 1st Floor\, Institute of Public Health\, Uni
 versity Forvie Site\, Robinson Way\, Cambridge
END:VEVENT
END:VCALENDAR