3inar / nowaclean Goto Github PK

Change to Mutual information: [number w 3 significant digits]

something like install_preprocessing() that does installs everything in suggested_packages from bioconductor

This is a lot faster than the built-in loess()

rewrite what the prob parameter does

increase it to 2 or something

This message re filtering out probes:

Filtered out 3417 probes.

Should maybe say why (aggregating across probes that map the same gene)

https://github.com/3inar/nowaclean/blob/master/R/preprocessing.R#L53

lumiHumanIDMapping

This message:

Removing 32748 probes that were not present in at least 25 % of the samples.

Should also take into acct that this is at a certain detection p-value threshold.

In preprocessing it would be nice to get an auto-generated bit of text that explains the process for use in papers and reports. Something like:

A probe must detect a signal with p-value <= [something] in at least [some fraction] of the samples. Removed [n1] probes that didn’t fulfill this. Removed an additional [n2] probes with bad annotation quality, finally filtered out [n3] probes by aggregating across probes that detect the same gene. This leaves [n] genes.

[ + more text here about background correction and normalization]