Magnitude of LFCs in f_madsTree

[TomHarrop]

At the moment extremely lowly-expressed genes will still be given a colour in the figure. Could remove is.na(padj)?

[TomHarrop]

Should I set a threshold on the LFC?

This will affect the number of genes that get padj = NA.

Don't know if it will affect the number of MADS genes; maybe for tomato?

• padj == NA means either:
  • all samples have zero counts
  • there was an extreme outlier
  • the row was filtered by automatic filtering (based on low mean normalized count)

Should I set an alpha threshold?

• automatic filtering is based on alpha. Lower alpha should mean more NAs.
  • for the rice dataset it seems like lowering alpha from 0.1 to 0.01 rejects an additional 1,824 genes

Check the number rejected from automatic filtering

dim(res[is.na(padj) & !is.na(pvalue)])[1]

• for default settings (lfcThreshold = 0, alpha = 0.1): it's 14,592
• lfcThreshold = 1: 18,240
• alpha = 0.01: 16,416
• alpha = 0.01, lfcThreshold = 1: 20,064

[TomHarrop]

Another option would be to set genes that weren't expressed in either sample to NA, but this would only work for rice.

[TomHarrop]

Or I could filter on p-value—only colour genes that are differentially expressed?

[TomHarrop]

Not a big deal if it's in SI. Sort-of-fixed in a20e48d2b6002a8d1c1382be62917126cde05391