SEQanswers

Go Back   SEQanswers > Bioinformatics > Bioinformatics



Similar Threads
Thread Thread Starter Forum Replies Last Post
DESeq2 Contrasts cacti Bioinformatics 4 08-04-2015 12:20 PM
DESeq2 Simon Anders Bioinformatics 123 07-06-2015 02:45 AM
Getting started with DESeq2 ThePresident Bioinformatics 9 02-27-2014 05:40 PM
DESeq2 SummarizedExperiment help sindrle Bioinformatics 10 10-23-2013 05:20 AM
Deseq : p-val & p-adj variation BUT foldchange CONSTANT chdem Bioinformatics 1 10-24-2012 08:09 AM

Reply
 
Thread Tools
Old 04-09-2014, 09:31 AM   #1
TheSeqGeek
Member
 
Location: USA

Join Date: Feb 2014
Posts: 40
Default DESeq2 p-adj or p-value

What is more accurate to use the padj or pvalue in RNA-seq?


Additionally, after my analysis I got a bunch NA for p-values.

Anyone know why this is the case or how to fix it?

Some have padj values but not pval and some both show NA for values.

Any thoughts?
TheSeqGeek is offline   Reply With Quote
Old 04-09-2014, 09:57 AM   #2
areyes
Senior Member
 
Location: Heidelberg

Join Date: Aug 2010
Posts: 165
Default

Hi @TheSeqGeek,

You have to use the adjusted p-values.

Everytime you do multiple tests (not only with RNA-seq data), you need to correct for what you would expect to be significant just by the number of tests you are performing. Say, if you tests 1000 hypothesis, you expect 100 (10%) of your data to have a p-value lower than 0.1. The adjusted p-values will already consider and correct for this.

The wikipedia page for "Multiple comparisons problem" can be of more help, the references from that page will point to relevant literature and methods for correcting for multiple testing.
areyes is offline   Reply With Quote
Old 04-09-2014, 10:21 AM   #3
Wallysb01
Senior Member
 
Location: San Francisco, CA

Join Date: Feb 2011
Posts: 286
Default

The “NA” issue is from not having enough read support to do a test. I believe the default is 10, but you can change that. This helps you because you aren’t going to be even attempting tests on lowly expressed genes, thereby lessing the impact of multiple testing corrections. In my samples, it seems about half of the genes aren’t even tested. So, with maybe ~22K genes or so, this means 11K tests to do multiple-test corrections on.
Wallysb01 is offline   Reply With Quote
Old 04-09-2014, 10:57 AM   #4
TheSeqGeek
Member
 
Location: USA

Join Date: Feb 2014
Posts: 40
Default

Thank you for the quick reply.

With regard to NA. Some of the samples that have NA have more than 100 mapped reads. Is the 10 referring to number of mapped reads or the difference between conditions?
TheSeqGeek is offline   Reply With Quote
Old 04-09-2014, 03:32 PM   #5
dpryan
Devon Ryan
 
Location: Freiburg, Germany

Join Date: Jul 2011
Posts: 3,480
Default

It's likely that many of the genes were either flagged as having outlier samples (see discussion of Cook's distance in the vignette) or were simply filtered during the automatic independent filtering step, which increases power.
dpryan is offline   Reply With Quote
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off




All times are GMT -8. The time now is 05:46 PM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2021, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO