Bioinformatics Seminar
Time:
Venue: Na
4 June 2019
NaAlways log
Terry SpeedWEHI Bioinformatics
I came to this view many years ago ; and found that it has served me well ; as so much of our data are intensities or counts. The work in progress which I will report was stimulated by data from some CRISPR/Cas9 screens carried out by Mingjie Luo. He did not get any biologically interesting results from the standard analysis of his data. In discussion ; we wondered whether that was because the data contained no biologically interesting results to be found ; or whether it did ; but the standard analysis failed to reveal them. Spoiler alert: I am not going to say which it is in my talk. Mingjie will do that in his masters completion seminar ; beginning in the same room an hour after mine ends (he hopes).
The story so far is a mix of theoretical speculation ; theory ; simulations ; and re-analysis of Mingjie's data. You can guess my role. Tony Pakes in UWA provided some theory. Mingjie did all the reanalysis and many of the simulations. We began by logging the read counts. Three surprises later ; we arrived at what seems to be a new and useful analysis of CRISPR/Cas9 screen data ; one which is very simple (always log!) and which seems to be valid under broader assumptions than are common in the literature. What is still to come is a thorough validation of the method using some of the ample published data with validated conclusions (positive controls). This talk is being given about 3 months too soon ; but the vacancy was on 4th June.;;;;