conflict with log2fc and raw counts

10 views
Skip to first unread message

mamta masand

unread,
Jul 29, 2019, 4:15:53 AM7/29/19
to trinityrnaseq-users
Hello everyone
I am analyzing DGE of control sample vs Treatment. After differential gene expression i got fold change for all contrasting genotypes when   I manually aligned the raw count values to transcripts then I found some conflicting values  there are some genes where raw count is higher in Treatment so it have to give negative fold change but it is giving positive fold change. Please suggest something.

Tiago Hori

unread,
Jul 29, 2019, 6:40:31 AM7/29/19
to mamta masand, trinityrnaseq-users
Don’t forget that most abundance estimators like RSEM internally normalize the data and also use likelihood to determine what mappings are correct. Counts do not necessarily reflect the data used for DE. Having said that, if the difference are high, you may have moose in your data. 

How are you “re-aligning” the reads?

T.

“If equal affection cannot be, let the more loving one be me” W.H Auden 

On Jul 29, 2019, at 5:15 AM, mamta masand <mamtama...@gmail.com> wrote:

Hello everyone
I am analyzing DGE of control sample vs Treatment. After differential gene expression i got fold change for all contrasting genotypes when   I manually aligned the raw count values to transcripts then I found some conflicting values  there are some genes where raw count is higher in Treatment so it have to give negative fold change but it is giving positive fold change. Please suggest something.

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/trinityrnaseq-users/9f0a67f7-8f38-4c5e-8390-e2603b0b62b1%40googlegroups.com.

Brian Haas

unread,
Jul 29, 2019, 9:20:28 AM7/29/19
to Tiago Hori, mamta masand, trinityrnaseq-users
The normalized counts (counts per million) will give you the fold change info - counts might be higher in one condition, but if it has total read counts that are also higher, that might explain the negative fold change.



--
--
Brian J. Haas
The Broad Institute
http://broadinstitute.org/~bhaas

 

mamta masand

unread,
Jul 31, 2019, 5:07:13 AM7/31/19
to trinityrnaseq-users
Hello everyone 
I have checked my files again and proceeded with analysis but I am facing the same problem again. I am attaching screenshot of foles where there is conflict between the values. Please suggest something.


On Monday, July 29, 2019 at 6:50:28 PM UTC+5:30, Brian Haas wrote:
The normalized counts (counts per million) will give you the fold change info - counts might be higher in one condition, but if it has total read counts that are also higher, that might explain the negative fold change.

On Mon, Jul 29, 2019 at 6:40 AM 'Tiago Hori' via trinityrnaseq-users <trinityrn...@googlegroups.com> wrote:
Don’t forget that most abundance estimators like RSEM internally normalize the data and also use likelihood to determine what mappings are correct. Counts do not necessarily reflect the data used for DE. Having said that, if the difference are high, you may have moose in your data. 

How are you “re-aligning” the reads?

T.

“If equal affection cannot be, let the more loving one be me” W.H Auden 

On Jul 29, 2019, at 5:15 AM, mamta masand <mamtama...@gmail.com> wrote:

Hello everyone
I am analyzing DGE of control sample vs Treatment. After differential gene expression i got fold change for all contrasting genotypes when   I manually aligned the raw count values to transcripts then I found some conflicting values  there are some genes where raw count is higher in Treatment so it have to give negative fold change but it is giving positive fold change. Please suggest something.

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-users+unsub...@googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-users+unsub...@googlegroups.com.
Screenshot (9).png

Tiago Hori

unread,
Jul 31, 2019, 8:01:35 AM7/31/19
to mamta masand, trinityrnaseq-users
Is it possible that your samples files is in the inverted order from what you expect (i.e your are expecting A/B, but it is calculating fold-changes B/A)? Can you send the samples file you used for DEG? 

T.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/trinityrnaseq-users/61e8289b-b4b5-425b-b473-908bfcc6fbbb%40googlegroups.com.

mamta masand

unread,
Jul 31, 2019, 8:29:10 AM7/31/19
to trinityrnaseq-users

I have attached the sample file but I used the command but the samples orientation is same. It is giving fold change accurate for other genes but in between it is giving different fold change.
samples.PE.txt

Brian Haas

unread,
Jul 31, 2019, 9:41:35 AM7/31/19
to mamta masand, trinityrnaseq-users
Hi,

This is a bit confusing.  Are you wanting to do a control vs. treatment comparison here?  Is the TMM value you're showing the mean value between replicates?

Note, if you use DESeq2 for doing the DE analysis, it will 'squeeze' fold changes for lowly expressed genes and so the fold changes you observe won't always match up to what you might calculate directly based on CPM values.

On Wed, Jul 31, 2019 at 8:29 AM mamta masand <mamtama...@gmail.com> wrote:

I have attached the sample file but I used the command but the samples orientation is same. It is giving fold change accurate for other genes but in between it is giving different fold change.
On Wednesday, July 31, 2019 at 5:31:35 PM UTC+5:30, Tiago Hori wrote:
Is it possible that your samples files is in the inverted order from what you expect (i.e your are expecting A/B, but it is calculating fold-changes B/A)? Can you send the samples file you used for DEG? 

T.

------ Original Message ------
From: "mamta masand" <mamtama...@gmail.com>
To: "trinityrnaseq-users" <trinityrn...@googlegroups.com>
Sent: 7/31/2019 6:07:13 AM
Subject: Re: [trinityrnaseq-users] conflict with log2fc and raw counts

Hello everyone 
I have checked my files again and proceeded with analysis but I am facing the same problem again. I am attaching screenshot of foles where there is conflict between the values. Please suggest something.

On Monday, July 29, 2019 at 6:50:28 PM UTC+5:30, Brian Haas wrote:
The normalized counts (counts per million) will give you the fold change info - counts might be higher in one condition, but if it has total read counts that are also higher, that might explain the negative fold change.

On Mon, Jul 29, 2019 at 6:40 AM 'Tiago Hori' via trinityrnaseq-users <trinityrn...@googlegroups.com> wrote:
Don’t forget that most abundance estimators like RSEM internally normalize the data and also use likelihood to determine what mappings are correct. Counts do not necessarily reflect the data used for DE. Having said that, if the difference are high, you may have moose in your data. 

How are you “re-aligning” the reads?

T.

“If equal affection cannot be, let the more loving one be me” W.H Auden 

On Jul 29, 2019, at 5:15 AM, mamta masand <mamtama...@gmail.com> wrote:

Hello everyone
I am analyzing DGE of control sample vs Treatment. After differential gene expression i got fold change for all contrasting genotypes when   I manually aligned the raw count values to transcripts then I found some conflicting values  there are some genes where raw count is higher in Treatment so it have to give negative fold change but it is giving positive fold change. Please suggest something.

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-u...@googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-u...@googlegroups.com.


--
--
Brian J. Haas
The Broad Institute
http://broadinstitute.org/~bhaas

 

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-u...@googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/trinityrnaseq-users/7334a25a-329d-4af9-beca-f6f44bc8d85a%40googlegroups.com.

mamta masand

unread,
Jul 31, 2019, 10:02:53 AM7/31/19
to trinityrnaseq-users

Iam using edgeR for DGE and I have analysed the data like control_vs_Treatment.
Hello everyone 
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-users+unsub...@googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-users+unsub...@googlegroups.com.


--
--
Brian J. Haas
The Broad Institute
http://broadinstitute.org/~bhaas

 

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-users+unsub...@googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-users+unsub...@googlegroups.com.

mamta masand

unread,
Jul 31, 2019, 10:31:39 AM7/31/19
to trinityrnaseq-users

As there is no replicates so i have analysed data with edgeR
Hello everyone 
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-users+unsub...@googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-users+unsub...@googlegroups.com.


--
--
Brian J. Haas
The Broad Institute
http://broadinstitute.org/~bhaas

 

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-users+unsub...@googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "trinityrnaseq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trinityrnaseq-users+unsub...@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages