Is logFC a log2 fold change or some other base in scran::findMarkers
1
1
Entering edit mode
@chris-conley-20973
Last seen 3.6 years ago
Huntsman Cancer Institute

Aaron, I searched ?scran::pairwiseTTests as well as the code on GitHub but could not find out whether the lfc argument and corresponding output column prefixed with logFC was log2 or natural log. I assume that it is base 2 as the limma output. Thanks in advance! Chris

scran • 689 views
2
Entering edit mode
Aaron Lun ★ 28k
@alun
Last seen 15 hours ago
The city by the bay

It's the same base as the log-transformation that was used to create the assay.type. If you were following the standard workflow using, e.g., scater::normalize, then yes, this would be base 2. If you did your own thing, then it's whatever base you used to compute the input x. I'll mention this explicitly in the documentation.

0
Entering edit mode

Thanks, I give a strong endorsement of the simple single cell workflow.