Question: assigning differential exons to trascripts
0
gravatar for Assa Yeroslaviz
3.2 years ago by
Assa Yeroslaviz1.4k
Munich, Germany
Assa Yeroslaviz1.4k wrote:

Hi all,

I would like to know, if there is a possibility to assign differential exons to specific transcripts.

I have run DEXSeq on my data set and have found (e.g.) one gene of interest with differential exon usage. This gene though has several transcripts (58 according to DEXSeq, biologically speaking still 14 different transcripts. s below).

I would like to know, if it is possible to assign a specific transcript to the differentiated exon(s), so that i can claim one transcript is higher expressed than another. After reading the DEXSeq information and also searching the internet, I assume it is not possible to do in DEXSeq. Therefore I would like to know, if there is a different way of getting this information.

I have also ran cuffQuant-cuffDiff on the bam files and have the count files. Is it a valid option to use the fold-change results from the cuffDiff run together with the DEXSeq results to claim such transcripts to be differentially expressed?

is there a different way of doing it?

 

thanks

Assa

ADD COMMENTlink modified 3.2 years ago by James W. MacDonald51k • written 3.2 years ago by Assa Yeroslaviz1.4k
Answer: assigning differential exons to trascripts
2
gravatar for James W. MacDonald
3.2 years ago by
United States
James W. MacDonald51k wrote:

RSEM, salmon, kallisto, cufflinks, and likely many other aligners are all intended to map reads to transcripts. Salmon and kallisto are much faster than the other two. Then you need to use something like EBseq to incorporate the uncertainty of your transcript counts into the comparisons. You could also use sleuth, which will take as input either kallisto data directly (or if you use Rob Patro's fork, salmon data).

ADD COMMENTlink written 3.2 years ago by James W. MacDonald51k

Thanks Jim for the fast response. 

I have thought about this problem already and have tried cuffDiff (w.o. cufflinks, as i am not interested in novel information). cuffDiff though find for the gene  shown above none of the transcripts to be significant. I am now trying the kallisto-sleuth method to see if i get anything different. I will also give EBSeq a go if I'll have the time.

I would still like to know, if there is a way to assign a transcript to the significant exons found by DEXSeq (@Alejandro Reyes)

ADD REPLYlink written 3.2 years ago by Assa Yeroslaviz1.4k
1

HI Assa, 

The slot 'mcols( dxd )$transcripts' contains a list of character vectors. In each slot you will find the transcript identifiers used to define an exon bin. 

Alejandro

ADD REPLYlink written 3.2 years ago by Alejandro Reyes1.7k

thanks.

So if I have (e.g.) two transcripts for a specific exon, it means this significant exon might be used by both of them. There is no way to know/differentiate between them .

Is it possible to run DEXSeq on an isoform (transcript) level?

ADD REPLYlink written 3.2 years ago by Assa Yeroslaviz1.4k
1

Hi Assa,

Yes, the 'default' DEXSeq counting scheme will focus on exons, without differentiation between the isoforms that originated this exon bins. So if an exon bin is differentially used, it could be an effect coming from any of the transcripts that overlap with that exon bin.

The DEXSeq implementation should handle transcript-level counts. Although I have not explore that option myself, I would suggest you have a look into this manuscript:

http://f1000research.com/articles/5-1356/v1

Alejandro

ADD REPLYlink written 3.2 years ago by Alejandro Reyes1.7k

Thanks, it sounds also promising. I'll read the paper and see how it works

ADD REPLYlink written 3.2 years ago by Assa Yeroslaviz1.4k

Thanks @James W. MacDonald for the response, but this is not exactly what I need.

I am not looking for a way to compare the two conditions against each other for a specific gene. With this i can use kallisto or deseq2.

I would like to know if there is a method to compare the expression (=read counts) of one transcript against a second transcript. 

as an example, let's say I am interested in the differences between the expression strength of the two STAT3 transcripts stat3a and stat3b (out of 14 different stat3 transcripts).

I would like to show that the expression changes of the stat3a transcript is stronger than the expression differences of the stat3b transcript when comparing one condition with another. It is something like a four-way comparison ( I hope it is clearer now).

 

ADD REPLYlink modified 3.2 years ago • written 3.2 years ago by Assa Yeroslaviz1.4k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 288 users visited in the last hour