created by Geraldine_VdAuwera
on 2015-10-25
Oxidation of guanine to 8-oxoguanine is one of the most common pre-adapter artifacts associated with genomic library preparation, arising from a combination of heat, shearing, and metal contaminates in a sample (doi: 10.1093/nar/gks1443). The 8-oxoguanine base can pair with either cytosine or adenine, ultimately leading to G→T transversion mutations during PCR amplification.
This occurs when a G on the template strand is oxidized, giving it an affinity for binding to A rather than the usual C. Thus, PCR will introduce apparent G>T substitutions in read 1 and C>A in read 2. In the resulting alignments, a given G>T or C>A observation could either be:
1. a true mutation
2. an 8-oxoguanine artifact
3. some other kind of artifact.
The variants (C→A)/(G→T) tend to occur in specific sequence contexts e.g. CCG→CAG (doi:10.1093/nar/gks1443). Although occurring at relatively low frequencies, these artifacts can have profound impacts on variant calling fidelity (doi:10.1093/nar/gks1443).
Updated on 2015-10-25
From miguelb on 2016-01-06
Hi Geraldine, We’ve noticed the very phenomenon you describe and allude to in the referenced paper in some of our custom capture sequencing experiments. While I do notice some tools in which you can flag a sample for having possible oxidative damage events (CollectSequencingArtifactMetrics and CollectOxoGMetrics in picard), is there one that I’ve missed in which you’d be able to flag a variant call from mutect as being a result of oxidative damage? Thanks, Miguel
From Geraldine_VdAuwera on 2016-01-06
Hi @miguelb,
I believe the Cancer Genome Analysis group here at Broad has some tools to do this; check out http://www.broadinstitute.org/cancer/cga/dtoxog