So here is a dumb question for someone like me who forgot her high school math. I've seen in papers that they cut by two fold a log2 fold change in RNA-seq data. So let's say that I have 5 up regulated genes like in the table above, if I cut by two fold, does it mean that I only keep log2 fold changes that are higher than 2?
Gene log2fold
A 5.00
B 4.00
C 3.00
D 1.55
E 1.00
1 answer
We generally use a two-fold cut-off on Fold Changes (FC) itself rather than log2(FC). Two-fold cut-off means anything that has a value greater than or equal to +2.0 and less than or equal to -2.0. This is because we need to find genes that are upregulated (positive FC) as well as downregulated (negative FC). Lets call the data that you have shown as dat. If you want to use this cut-off on this data, you would do something like this in R:
dat.log2FC.twofold = dat[abs(dat$log2fold) >= 2.0,]
This will give you log2 fold changes values greater than or equal to +2.0 and less than or equal to -2.0. It is just a matter of taking an absolute of your value and applying the cut-off on it, rather than simply applying the cut-off.
•
written
by
komal.rathi
Log in to answer this question.
-
Before doing pseudobulk, should we get rid of genes that are only expressed in a small number of ce…
written by mropri
-
R function for subsetting rows from one list based on certain conditions met from that list and ano…
written by jamzaleg84 •
-
Loop function for ggplot2 bar chart for multiple variants
written by luongthang1908
-
Forum: Potential post-doc PI giving exam as part of hiring process
written by MAPK
-
Fold change cut off enrichment analysis
written by Pietro
-
General questions about running GSEAPreranked
written by mmccarthy781 •
-
Comparison of two MeDip data sets
written by florian.noack •
-
Question about the DESeq2: log2foldchange contradictory
written by tiago211287
-
Primer3 Output File Format
written by najoshi •
-
Fold Change Differences Between Mbni'S Custom Cdf Probeset And Affymetrix Original Probesets (Netaf…
written by Lei Huang •
Just as a side comment, ranking genes and setting a cut-off on logFC tends to include mostly genes expressed at low levels. This is because statistically and maybe also biologically, low expressed genes tend to have larger differences.
Conversely, ranking and filtering by FDR tends to favour genes expressed at high levels.
A workaround this is to select genes that are towards the perifery of the cloud of the MAplot (i.e. plot of expression vs logFC). I wrote a little R script to do this (intensityFilter.R).