First publication!

It has been a long time since my last blog post, but the news is that I published my first paper! It’s been a year and a half since I joined the Brent Lab and I have seen my rotation project project “Predicting which genes will respond to transcription factor perturbations” come to fruition.

Briefly, the paper aims to predict which genes will respond to the perturbation of a particular transcription factor (TF) in yeast and human cells. We use XGBoost gradient-boosted trees integrating pre-perturbation gene expression level and variation, histone modifications, chromatin accessibility, and ChIP-seq TF-binding location data to predict whether a gene is “responsive” to a TF perturbation. In human cells, a gene is considered responsive based on whether its DESeq2 result are significant at P<0.05 and its absolute log2 fold-change is greater than 0.5. One of the core analyses of the paper is the quantification of feature influences. That is, determining which of the input features is the most predictive. The paper finds that in humans, pre-perturbation gene expression level and variation were the most influential whereas TF binding location was least informative. This comes at a surprise as ChIP-seq data is often regarded as one of the most useful sources of evidence for potential TF regulation of a gene.

The lack of TF binding location predictive power in human cells is interesting, particularly because it is one of the most useful features in the yeast model. One way we could attribute this result is the size and complexity of the human genome. This may be evidence that in humans, TF regulation may be more mediated by enhancers that are further from the gene body.

Check out the paper published on G3 at the following link:

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: