Genomics in the Cloud

Chapter 7. GATK Best Practices for Somatic Variant Discovery

In this chapter, we tackle somatic variant discovery, mainly as it applies to cancer research. This is going to introduce new challenges and, correspondingly, new experimental and computational designs. We begin with somatic short variants, which still have a lot in common with germline short variants, with enough new challenges to keep things interesting. Then we expand our horizons to include copy-number variation, which involves a very different approach from what we’ve taken so far.

Challenges in Cancer Genomics

Before we get into the weeds, let’s make one thing clear: everything is more difficult with respect to cancer. Consider what is happening in the body before a tumor even begins to develop. Living cells in our bodies are not static; they are metabolically active, taking in nutrients, doing work, and pumping out waste. Some of them are dividing at various rates. Enzymes are unraveling DNA to transcribe it and/or copy and repackage it. At the scale of a single cell, very few of those molecular transactions ever go wrong, but because this is happening all the time in millions of cells, the numbers add up to a lot of little things going wrong, causing mutations across the board.

Most of the time, these mutations have no discernible effect whatsoever. But from time to time, a mutation arises that destabilizes the affected cell’s metabolism in a way that causes it to begin dividing more rapidly and produce a tumor. ...