This week we profile a recent publication in bioRxiv from Marjan Farahbod (pictured)
in the laboratory of Dr. Paul Pavlidis at the Michael Smith Laboratories.
Can you provide a brief overview of your lab’s current research focus?
The Pavlidis lab does computational biology and bioinformatics research. We’re primarily interested in studying the nervous system, using genomics and genetics data to help understand human conditions like schizophrenia and autism. But we also work on more generally-applicable questions, like how to discover gene networks and use them to learn about cell function, as in this preprint.
What is the significance of the findings in this publication?
A big goal of genomics is to try to understand how genes are regulated, that is, how a cell controls when a gene is “on” or “off”. One idea is that if some genes show a similar pattern of expression across conditions or samples – this is the “coexpression” of the title – they must be coregulated, and that this can tell you something about the network controlling those genes. It’s a very popular method, but it’s hard to point at specific regulation discovered this way.
What we show in the preprint is that coexpression doesn’t have a lot to do with regulation, at least not at a fine granularity. Instead, it tells you about which genes are expressed in which cell types, which is relatively boring. The idea that this is happening is not entirely new, but most users of coexpression don’t know about it. Or if they do, they don’t realize that it’s the dominant signal.
What are the next steps for this research?
In the preprint we compared coexpression between single-cell data and brain tissue, and like many people we’re interested in how single-cell data can help get at regulation. We also discuss the possibility of correcting tissue-level data for the cell-type effects, and we’re following up on that too.
This work was funded by:
The research in the paper was supported by grants from the US NIH and NSERC.