Tophat, Cufflinks and replicates - Page 2

RockChalkJayhawk · 12-06-2010, 02:59 PM

Quote:

Originally Posted by adarob

By "newest version" do you mean 0.9.2? There was a small bug in that release which in some cases caused some of the alignments to be ignored. If it is not too much trouble, could you rerun with 0.9.3 (released yesterday)?

This figure was generated using the same protocol as above.
Using v0.9.3 really made things better. Now I have a FDR of 2.13% (427/20014), much better than the 30% I had previously. There is still a bit more variation than from the counting methods, but at least I can work with this.

Thanks for the updates!

lahoman · 02-07-2011, 09:31 AM

Hi, xinchen,

Once I get the sam file using TopHat, how should be we count the reads for each gene? You know, I am new to RNA-Seq analysis. My question might be naive

Thank you so much,

Lahoman

nkwuji · 02-08-2011, 02:23 AM

Quote:

Originally Posted by lahoman

Hi, xinchen,

Once I get the sam file using TopHat, how should be we count the reads for each gene? You know, I am new to RNA-Seq analysis. My question might be naive

Thank you so much,

Lahoman

Hi, Lahoman,

Here is the introduction for cufflinks.
http://cufflinks.cbcb.umd.edu/manual.html

It is better to read through it. I thought there might be some shortcuts or easy ways to use it. But later on, I found it really necessary to read all the instructions carefully, and make sure about the parameters you should use.

Cheers,
Jun

lahoman · 02-08-2011, 12:47 PM

Hi,RockChalkJayhawk,

After I use tophat to map Human RNA-Seq to the genome, then cufflinks for the transcript analysis, I checked the file of transcripts.expr. There are
263,506 transcripts. That's a lot. Do I need to filter the results based on FPKM? What criteria should I use? 1.0 or 0.5? I have no idea about it.

Thanks,

Lahoman

RockChalkJayhawk · 02-08-2011, 01:24 PM

Quote:

Originally Posted by lahoman

Hi,RockChalkJayhawk,

After I use tophat to map Human RNA-Seq to the genome, then cufflinks for the transcript analysis, I checked the file of transcripts.expr. There are
263,506 transcripts. That's a lot. Do I need to filter the results based on FPKM? What criteria should I use? 1.0 or 0.5? I have no idea about it.

Thanks,

Lahoman

Unfortunately, there is no right answer. Everything is subjective. Just use good judgement...

chrisbala · 02-09-2011, 05:58 AM

I think this issue of excessive cufflinks predicted transcripts is one many people are wrestling with. I have not figured out a satisfactory way to deal with this other than just using ONLY existing Ensembl gene models (boo!). My interpretation of the problem is that pre-mRNA reads (and perhaps other forms of "background") give cufflinks trouble (understandably). I don't know if using paired-end data (which I did not do) and 100s of millions of reads (like the cufflinks paper did) as opposed to 10s of millions of read would help. If others have thoughts I would love to hear them..

lpachter · 02-10-2011, 05:15 AM

The reason for excessive transcripts is that in genes with low coverage, Cufflinks will likely predict multiple disjoint transcripts- that is in fact the only thing it can do because it is not a gene finder and has no reason to glue them together (its also true that "background" causes problems and makes assembly difficult).

To address these issues, and also the fact that many users want to discover novel isoforms with respect to a known annotation (e.g. ENSEMBL), we have developed a new reference annotation guided assembly mode that will take as input your annotation and assemble with respect to that. It results in much cleaner assemblies by merging partial transcript fragments into the known annotations, yet is able to reveal novel isoforms. It sounds like it addresses the issue (and desired feature) you are reporting on.

It will be released in the next week or two as soon as we finish some more testing, so please check back with us before the end of the month.

lahoman · 02-10-2011, 07:46 AM

Hi,
I checked the accepted_hits.sam generated by TopHat. For one reads pair, their mapping results are:

USI-EAS376_0001:3:112:7215:15873#0 129 chr1 14790 255 40M140N35M = 155253852 0 CCGGGCCCCTCACCAGCCCCAGGTCCTTTCCCAGAGATGCCCTTGCGCCTCATGACCAGCTTGTTGAAGAGATCC bbbbbabbbbbbbbbbbbbbbababbbbbbbb^bbbbcbbbb`cR`\aaY`\a`bbbcbb_\_^baa[`Wa]X_a NM:i:0 XS:A:- NS:i:0

The CIGAR code is 40M140N35M. We can see it spans two exons. 14790 on chr1 is the position in the first exon. How can we know the information about the second exon?

Thanks a lot,

Lahoman

lahoman · 02-10-2011, 07:58 AM

Hi,

Suddenly I realize that the second exon is from the same chromosome. The distance between the first and second exon is 140bp.
The one more question is, if a RNA-Seq reads is spliced and aligned to different chromsome, TopHat won't report it. Am I right? In other words, TopHat can't be used for genefusion analysis.

Thanks,

Lahoman

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Cufflinks, Cuffdiff, without replicates	brachysclereid	Bioinformatics	1	10-16-2012 03:43 AM
several questions about tophat and cufflinks	liuxq	Bioinformatics	0	11-15-2011 12:03 AM
DESeq: question about with replicates and without any replicates.	nb509	RNA Sequencing	2	10-25-2011 07:04 AM
Differential gene expression: Can Cufflinks/Cuffcompare handle biological replicates?	marcora	Bioinformatics	38	12-14-2010 04:57 PM
Differential gene expression: Can Cufflinks/Cuffcompare handle biological replicates?	marcora	Bioinformatics	0	05-19-2010 02:11 AM

02-07-2011, 09:31 AM	#22
lahoman Member Location: Houston Join Date: Jan 2011 Posts: 12	How did you count the reads from the assembly file? Hi, xinchen, Once I get the sam file using TopHat, how should be we count the reads for each gene? You know, I am new to RNA-Seq analysis. My question might be naive Thank you so much, Lahoman

02-08-2011, 12:47 PM	#24
lahoman Member Location: Houston Join Date: Jan 2011 Posts: 12	Cufflink results filtering? Hi,RockChalkJayhawk, After I use tophat to map Human RNA-Seq to the genome, then cufflinks for the transcript analysis, I checked the file of transcripts.expr. There are 263,506 transcripts. That's a lot. Do I need to filter the results based on FPKM? What criteria should I use? 1.0 or 0.5? I have no idea about it. Thanks, Lahoman

02-10-2011, 07:46 AM	#28
lahoman Member Location: Houston Join Date: Jan 2011 Posts: 12	the mapping position of RNA-Seq spanning junction Hi, I checked the accepted_hits.sam generated by TopHat. For one reads pair, their mapping results are: USI-EAS376_0001:3:112:7215:15873#0 129 chr1 14790 255 40M140N35M = 155253852 0 CCGGGCCCCTCACCAGCCCCAGGTCCTTTCCCAGAGATGCCCTTGCGCCTCATGACCAGCTTGTTGAAGAGATCC bbbbbabbbbbbbbbbbbbbbababbbbbbbb^bbbbcbbbb`cR`\aaY`\a`bbbcbb_\_^baa[`Wa]X_a NM:i:0 XS:A:- NS:i:0 The CIGAR code is 40M140N35M. We can see it spans two exons. 14790 on chr1 is the position in the first exon. How can we know the information about the second exon? Thanks a lot, Lahoman

02-09-2011, 05:58 AM	#26
chrisbala Member Location: North Carolina Join Date: Jan 2010 Posts: 82	I think this issue of excessive cufflinks predicted transcripts is one many people are wrestling with. I have not figured out a satisfactory way to deal with this other than just using ONLY existing Ensembl gene models (boo!). My interpretation of the problem is that pre-mRNA reads (and perhaps other forms of "background") give cufflinks trouble (understandably). I don't know if using paired-end data (which I did not do) and 100s of millions of reads (like the cufflinks paper did) as opposed to 10s of millions of read would help. If others have thoughts I would love to hear them..

02-10-2011, 05:15 AM	#27
lpachter Member Location: Berkeley, cA Join Date: Feb 2010 Posts: 40	The reason for excessive transcripts is that in genes with low coverage, Cufflinks will likely predict multiple disjoint transcripts- that is in fact the only thing it can do because it is not a gene finder and has no reason to glue them together (its also true that "background" causes problems and makes assembly difficult). To address these issues, and also the fact that many users want to discover novel isoforms with respect to a known annotation (e.g. ENSEMBL), we have developed a new reference annotation guided assembly mode that will take as input your annotation and assemble with respect to that. It results in much cleaner assemblies by merging partial transcript fragments into the known annotations, yet is able to reveal novel isoforms. It sounds like it addresses the issue (and desired feature) you are reporting on. It will be released in the next week or two as soon as we finish some more testing, so please check back with us before the end of the month.

02-10-2011, 07:58 AM	#29
lahoman Member Location: Houston Join Date: Jan 2011 Posts: 12	Hi, Suddenly I realize that the second exon is from the same chromosome. The distance between the first and second exon is 140bp. The one more question is, if a RNA-Seq reads is spliced and aligned to different chromsome, TopHat won't report it. Am I right? In other words, TopHat can't be used for genefusion analysis. Thanks, Lahoman