Question: Clustering genes in time course analysis according slopes in quadratic regression model
0
gravatar for antgomo
5.1 years ago by
antgomo30
Spain
antgomo30 wrote:

I have a RNA time course experiment, 7 time points, 2 replicates per point. After analyzed with DESeq2 package, I want to subset my results in 4 categories

1. Those genes that are increasing expression lineally from time 0 to 7

2. Those genes that are decreasing expression lineally from time 0 to 7

3. Those genes that increase expression from 0 to 3/4 and then decrease

4. Those genes that decrease expression from 0 to 3/4 and then increase

 

So I fitted a quadratic regression model

      results<-apply(expression.matrix,1,function(x) lm(y~as.numeric(x)+I(as.numeric(x)^2)))

where y is a vector comprising time from 1 to 7 

Now I extract x and x2 from the results

   expression.matrix$x2<-unlist(lapply(results, function(x) return(x$coef[3])))
    expression.matrix$x<-unlist(lapply(results, function(x) return(x$coef[2])))

For class 1 ad 2 above, I know that with x >0 and x<0, I can get both, but for 3 and 4 , I don't know i have to take ( in case 3), x=0 and x2>0 and x=0 and x2<0 for 4. The problem is, that I don't have values of x of o, and even if got those ones closer to 0, the curve is not behaving as I expected

 

Any ideas?

 

Thanks in advance
 

ADD COMMENTlink modified 5.0 years ago by Biostar ♦♦ 20 • written 5.1 years ago by antgomo30

I would suggest fuzzy clustering with the mfuzz package in R.

ADD REPLYlink written 5.1 years ago by Benn8.1k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1999 users visited in the last hour
_