range of fold change in RNA seq
1
2
Entering edit mode
7.1 years ago
kanwarjag ★ 1.2k

I performed an RNAseq experiment and used DESEq2 to calculate statistical significance of differential expression. When I calculated fold change it ranges from 4 fold to 700 fold. I randomly checked few genes showing very high fold change and observed that mostly all the replicates (three) showed good correlation in raw reads in each group and the difference looks reasonably real. However, I dont think any one can show such large fold change in publication or in analysis. . But biologist would like to show FC. I thought about showing log2 ratio but in other comparisons I am showing FC and can not make exception for one comparison. Any suggestion how to tackle such suggestion.

RNA-Seq • 4.1k views
ADD COMMENT
1
Entering edit mode

the difference looks reasonably real

I think a 700 fold change is quite suspicious... What is the biological setup?

ADD REPLY
0
Entering edit mode

Cells were treated with drug for 5 days in a time course experiment.

I think reviewers in publication will think it is an artifact, however since such large FC is not in other group comparisons in the same experiment provide an assurance that it real. It is challenging to present such large FC Thanks

ADD REPLY
1
Entering edit mode

I think reviewers in publication will think it is an artifact

Can't blame them...
Have you considered validating? Restarting the analysis from the beginning?

What are the raw counts you get for this extreme gene?

ADD REPLY
0
Entering edit mode

I have checked it very thoroughly there is no mistake. Here is snap shot of few such genes raw read from Deseq.

Gene    grp1    grp1    grp1    grp2    grp2    grp2
A1  1540    1438    1596    2   4   4
A2  2422    2426    2452    28  20  16
A3  1318    1408    1520    8   18  14
A4  2076    2180    2392    6   14  4
A5  994 994 1102    6   4   10
A6  1354    1444    1552    8   14  12
A7  2456    2572    2674    8   14  18
A8  936 1108    1058    12  10  6
A9  1614    1708    1854    8   10  10

First three columns one grp and last three columns second grp. grp1 is treated and induced expression grp2 is untreated control

ADD REPLY
0
Entering edit mode

I don't see any columns...

And please, do not add further information as an answer, update you original post.

ADD REPLY
0
Entering edit mode

I added markup to your post for increased readability. You can do this by selecting the text and clicking the 101010 button. When you compose or edit a post that button is in your toolbar, see image below:

101010 Button

Please use ADD COMMENT or ADD REPLY to answer to previous reactions, as such this thread remains logically structured and easy to follow. I have now moved your post but as you can see it's not optimal. Adding an answer should only be used for providing a solution to the question asked.

ADD REPLY
0
Entering edit mode

Is grp1 treated or control? i.e did the drug induce expression or repress?

ADD REPLY
0
Entering edit mode

grp1 is treated induced expression grp2 is untreated control

ADD REPLY
1
Entering edit mode

Have they been sequenced at the same time? Are the coverages roughly similar (size factors?) ?

ADD REPLY
0
Entering edit mode

Post the normalized counts. In general, though, if the genes are going from "little/no expression" to "moderate/high expression" and this seems biologically reasonable given whatever the drug is expected to do then there's little reason to complain. You should certainly validate a few of those regardless.

ADD REPLY
0
Entering edit mode
Gene grp1   grp1    grp1    grp2    grp2    grp2
    1   5.30    5.22    5.14    -3.82   -4.75   -4.49
    2   4.87    4.74    4.87    -4.82   -4.75   -4.49
    3   6.51    6.24    6.55    -3.24   -1.94   -2.91
    4   4.93    4.66    4.83    -4.82   -3.75   -3.49

normalized counts of 4 such genes Grp 1 is treated Grp2 is control

ADD REPLY
0
Entering edit mode

How exactly are you getting negative normalized counts? Are these log2 transformed or something?

ADD REPLY
0
Entering edit mode
7.1 years ago
theobroma22 ★ 1.2k

I've gotten FC in the thousands. It's not uncommon. For example, in plants Ethylene and Jasmonic Acid can have very large FCs depending on the experimental design. The question is what are those high FC gene models? Further, including them is based on your hypothesis.

ADD COMMENT
0
Entering edit mode

Did you publish such large FC, any link for your or any such publications. Generally a FC of <100 is reported. Were reviewers OK with such FC?

ADD REPLY
0
Entering edit mode

Thousands can be acceptable. Tens of thousands is usually scrutinized.

ADD REPLY

Login before adding your answer.

Traffic: 1496 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6