A major part of bioinformatic endeavor comprises statistical analysis, and it is in our best interest to understand the ins and outs of the models we employ, to know why they are used, what they are good for, and most importantly where they fall short while seeking to answer a specific scientific question.

Negative binomial distribution is the poster child distribution for RNA-seq studies because the measured variable is in the unit of counts. Packages such as edgeR and DESeq are widely used by the RNA-seq community and their underlying model is based on this model. However, the manuals and journals usually do not go into details of the models being employed and often assume that the reader is familiar with the very concepts required to understand their idea.

In this an hour-long informal seminar, I will derive the negative binomial distribution from the ground up using plain English and first principles, showing the step-by-step derivation from the premise of RNA-seq. I gave this talk during our lab weekly journal club a while ago and also to a couple of my colleagues, and it was well received and I want to share this knowledge to the wider community.

Pizza and soda will be served.