Identify and visualize differentially expressed genes from RNA-Seq data?

mediator · 02-01-2012, 09:57 AM

Hi All,
I have aligned RNA Seq data for four healthy controls and four patients, in bam format. Then I made a table for RPKM for all the genes. Now how do I quickly identify and visualize those differentially expressed genes in patients? In other words, how to identify which genes are expressed more in patients but not in healthy control? I've tried SAM (significant analysis for microarray ) but no good. Any suggestions? Thanks!

chadn737 · 02-01-2012, 10:03 AM

There are several programs in R that are for differential expression of RNA-seq data:

EdgeR
DESeq
BaySeq
DEGseq

However the input for these programs is raw read counts, not RPKM normalized genes.

Richard Finney · 02-01-2012, 10:08 AM

What tool did you use?
is the result by exon? or gene? how does it correct for gene overlap?

How would you like to visualize the results? As a list? On the genome? On pathways or ontologies?

mediator · 02-01-2012, 10:12 AM

@chadn737
Thanks!
@Richard Finney
I used Seqmonk to generate RPKM , results are by exons but can be averaged to by genes. I would like to visualize in a hierarchical clustered heat map, like many microarray data.

kjlee · 03-23-2012, 12:08 PM

Hello mediator,

I have found that MeV is very helpful in visualizing RNA-seq expression data. It is well-documented and easy to implement.

Cheers,
Kevin

mediator · 03-25-2012, 11:51 AM

@kjlee
Thanks for the advice! I already use MeV for microarray. Pretty handy. Let's see how RNA-Seq data goes with MeV.

kumardeep · 05-09-2012, 12:02 AM

Hi all
I have RNA-seq data (2 control sets and 2 test sets). I have generated the *.sorted.bam files (~1GB) through bowtie software. Right now I am using cufflinks software to generate RPKM values to get differential gene expression quantitatively. In cufflinks, I followed the command:
cufflinks control_1.sorted.bam
It takes much time on cluster (thats reasonable for 1gb INPUT files I guess). The output file (say transcript.gtf) is having RPKM values. I want to know how to proceed further to get final list of differentially expressed genes with corresponding RPKM values.
Do I have to run the above command for all input files (viz. control_2.sorted.bam, test_1.sorted.bam, test_2.sorted.bam)?
Also How to compare RPKM values finally ?
Thanks in advance..
Kumardeep

LiamT · 05-09-2012, 01:00 AM

Hi Kumardeep

You should always start a new thread if you're asking a new question, even if it is related (you can always refer to the other thread).

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Cufflinks, differentially expressed genes	statsteam	Bioinformatics	5	11-15-2013 12:28 PM
Getting differentially expressed genes based on RPKM values	casshyr	RNA Sequencing	3	04-30-2012 10:18 AM
DESeq and EdgeR: too many differentially expressed genes!?!?	cutcopy11	Bioinformatics	5	12-08-2011 01:14 AM
Comparing mouse and human differentially expressed genes	stephenhart	General	3	11-16-2011 02:14 AM
RNA-Seq: Proportion statistics to detect differentially expressed genes: a comparison	Newsbot!	Literature Watch	0	06-09-2011 03:00 AM

02-01-2012, 09:57 AM	#1
mediator Member Location: New England Join Date: Nov 2010 Posts: 27	Identify and visualize differentially expressed genes from RNA-Seq data? Hi All, I have aligned RNA Seq data for four healthy controls and four patients, in bam format. Then I made a table for RPKM for all the genes. Now how do I quickly identify and visualize those differentially expressed genes in patients? In other words, how to identify which genes are expressed more in patients but not in healthy control? I've tried SAM (significant analysis for microarray ) but no good. Any suggestions? Thanks!

02-01-2012, 10:03 AM	#2
chadn737 Senior Member Location: US Join Date: Jan 2009 Posts: 392	There are several programs in R that are for differential expression of RNA-seq data: EdgeR DESeq BaySeq DEGseq However the input for these programs is raw read counts, not RPKM normalized genes.

02-01-2012, 10:08 AM	#3
Richard Finney Senior Member Location: bethesda Join Date: Feb 2009 Posts: 700	What tool did you use? is the result by exon? or gene? how does it correct for gene overlap? How would you like to visualize the results? As a list? On the genome? On pathways or ontologies?

02-01-2012, 10:12 AM	#4
mediator Member Location: New England Join Date: Nov 2010 Posts: 27	@chadn737 Thanks! @Richard Finney I used Seqmonk to generate RPKM , results are by exons but can be averaged to by genes. I would like to visualize in a hierarchical clustered heat map, like many microarray data.

03-23-2012, 12:08 PM	#5
kjlee Member Location: Atlanta, GA Join Date: Jun 2011 Posts: 12	Hello mediator, I have found that MeV is very helpful in visualizing RNA-seq expression data. It is well-documented and easy to implement. Cheers, Kevin

03-25-2012, 11:51 AM	#6
mediator Member Location: New England Join Date: Nov 2010 Posts: 27	@kjlee Thanks for the advice! I already use MeV for microarray. Pretty handy. Let's see how RNA-Seq data goes with MeV.

05-09-2012, 12:02 AM	#7
kumardeep Member Location: Chandigarh, India Join Date: Apr 2012 Posts: 15	Hi all I have RNA-seq data (2 control sets and 2 test sets). I have generated the *.sorted.bam files (~1GB) through bowtie software. Right now I am using cufflinks software to generate RPKM values to get differential gene expression quantitatively. In cufflinks, I followed the command: cufflinks control_1.sorted.bam It takes much time on cluster (thats reasonable for 1gb INPUT files I guess). The output file (say transcript.gtf) is having RPKM values. I want to know how to proceed further to get final list of differentially expressed genes with corresponding RPKM values. Do I have to run the above command for all input files (viz. control_2.sorted.bam, test_1.sorted.bam, test_2.sorted.bam)? Also How to compare RPKM values finally ? Thanks in advance.. Kumardeep

05-09-2012, 01:00 AM	#8
LiamT Junior Member Location: South Africa Join Date: Apr 2012 Posts: 5	Hi Kumardeep You should always start a new thread if you're asking a new question, even if it is related (you can always refer to the other thread).