Best pipeline for DE and fusion gene using tophat?

[ymc]

I have rna-seq data that I want to perform DE and find fusion genes. I am more into accuracy than speed. My reads are about 78bp long. Can you tell me which pipeline is better? Or is there an even better way to do this?

1) bowtie2 only pipeline
===============

1. Run tophat2 --fusion-search with bowtie2 for two samples.

2. Run tophat-fusion-post on the two outputs for fusion genes.

3. Run cuffdiff on the two outputs for DE.

2) bowtie2 for DE and bowtie1 for fusion
=========================

1. Run tophat2 --fusion-search --bowtie1 for two samples

2. Run tophat-fusion-post on the two outputs for fusion genes.

3. Run tophat2 with bowtie2 for two samples.

4. Run cuffdiff for the two outputs.

Which pipeline will provide more accurate result? Thanks!

[ndaniel]

It has been forgotten the options 3 and 4 which are

3) use SOAPFuse http://soap.genomics.org.cn/soapfuse.html for finding fusion genes

or

4) use FusionCatcher https://code.google.com/p/fusioncatcher/ for finding fusion genes

TopHat-Fusion has quite high false-positive rate regarding fusion genes compared with other fusion finder tools (for example SOAPfuse).
I would say the answer to your question is option 3 or option 4.

I have rna-seq data that I want to perform DE and find fusion genes. I am more into accuracy than speed. My reads are about 78bp long. Can you tell me which pipeline is better? Or is there an even better way to do this?

1) bowtie2 only pipeline
===============

1. Run tophat2 --fusion-search with bowtie2 for two samples.

2. Run tophat-fusion-post on the two outputs for fusion genes.

3. Run cuffdiff on the two outputs for DE.

2) bowtie2 for DE and bowtie1 for fusion
=========================

1. Run tophat2 --fusion-search --bowtie1 for two samples

2. Run tophat-fusion-post on the two outputs for fusion genes.

3. Run tophat2 with bowtie2 for two samples.

4. Run cuffdiff for the two outputs.

Which pipeline will provide more accurate result? Thanks!

[jp.]

Dear Senior Member
I am asking a help in tophat fusion. I have run this successfully with hg19. but got error with mm10.
Q1. BLAST Database error: No alias or index file found for nucleotide database [blast/nt] in search path
Q2. However, I continued it took long time compare to normal process in fusion-post and got fusions but the the other samples gives 0 fusion within 10 seconds. why ?
I have posted here: http://seqanswers.com/forums/showthr...613#post112613
I would really appreciate if you may answer please ..

I have rna-seq data that I want to perform DE and find fusion genes. I am more into accuracy than speed. My reads are about 78bp long. Can you tell me which pipeline is better? Or is there an even better way to do this?

1) bowtie2 only pipeline
===============

1. Run tophat2 --fusion-search with bowtie2 for two samples.

2. Run tophat-fusion-post on the two outputs for fusion genes.

3. Run cuffdiff on the two outputs for DE.

2) bowtie2 for DE and bowtie1 for fusion
=========================

1. Run tophat2 --fusion-search --bowtie1 for two samples

2. Run tophat-fusion-post on the two outputs for fusion genes.

3. Run tophat2 with bowtie2 for two samples.

4. Run cuffdiff for the two outputs.

Which pipeline will provide more accurate result? Thanks!

[ymc]

Dear Senior Member
I am asking a help in tophat fusion. I have run this successfully with hg19. but got error with mm10.
Q1. BLAST Database error: No alias or index file found for nucleotide database [blast/nt] in search path
Q2. However, I continued it took long time compare to normal process in fusion-post and got fusions but the the other samples gives 0 fusion within 10 seconds. why ?
I have posted here: http://seqanswers.com/forums/showthr...613#post112613
I would really appreciate if you may answer please ..

For your Q1, did you download mouse blast db file?

[jp.]

I have downloaded blast as follows from ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/:
1. ncbi-blast-2.2.28+
2. extracted est_mouse.tar.gz mouse_genomic_transcript.tar.gz within ncbi-blast-2.2.28+/ and ncbi-blast-2.2.28+/bin
I am not sure about other_genomic* and nt* ?
I used and exported PATH=$PATH:bowtie1, tophat2, blast, samtools
I have all the index of Mus_musculus refseq from UCSC, Ensembl.
Am I missing something ?

For your Q1, did you download mouse blast db file?

[ymc]

I don't see any file related to mouse here

ftp://ftp.ncbi.nlm.nih.gov/blast/db/

but according to the manual, it didn't say whether other organisms need these file. It only said you need to add --non-human option

[jp.]

Yes, I used --non-human option but my fusion results are same without difference with or without --non-human option.
I used recommended reference files.
1. However, the error was there and it gave me fusions, when I continued with errors.
2. for the other samples, it is giving no error but 0 fusions.

Can you please guess something ?

I don't see any file related to mouse here

ftp://ftp.ncbi.nlm.nih.gov/blast/db/

but according to the manual, it didn't say whether other organisms need these file. It only said you need to add --non-human option

[ymc]

Yes, I used --non-human option but my fusion results are same without difference with or without --non-human option.
I used recommended reference files.
1. However, the error was there and it gave me fusions, when I continued with errors.
2. for the other samples, it is giving no error but 0 fusions.

Can you please guess something ?

I am not doing anything related to non-human, so I am afraid I am not the right person for you to ask. Maybe you should send an email to tophat authors directly? Or try other tools that have better documentation about non-human applications.

Good luck!

[ymc]

I am curious why do you need to find fusion genes in mouse? Are you studying cancer specific to mouse?

If you are working on human cancer xenograft, then I suppose the rna-seq data you generated is still human not mouse.

[jp.]

These are from the cancer models, therefore I need to do it.
What you think is the reason of the problem with blast /nt?

I am curious why do you need to find fusion genes in mouse? Are you studying cancer specific to mouse?

If you are working on human cancer xenograft, then I suppose the rna-seq data you generated is still human not mouse.

[ymc]

Did you download all these files?

nt.00.nhd nt.02.nhr nt.04.nnd nt.06.nog nt.08.nsi nt.11.nhd nt.13.nhr
nt.00.nhi nt.02.nin nt.04.nni nt.06.nsd nt.08.nsq nt.11.nhi nt.13.nin
nt.00.nhr nt.02.nnd nt.04.nog nt.06.nsi nt.09.nhd nt.11.nhr nt.13.nnd
nt.00.nin nt.02.nni nt.04.nsd nt.06.nsq nt.09.nhi nt.11.nin nt.13.nni
nt.00.nnd nt.02.nog nt.04.nsi nt.07.nhd nt.09.nhr nt.11.nnd nt.13.nog
nt.00.nni nt.02.nsd nt.04.nsq nt.07.nhi nt.09.nin nt.11.nni nt.13.nsd
nt.00.nog nt.02.nsi nt.05.nhd nt.07.nhr nt.09.nnd nt.11.nog nt.13.nsi
nt.00.nsd nt.02.nsq nt.05.nhi nt.07.nin nt.09.nni nt.11.nsd nt.13.nsq
nt.00.nsi nt.03.nhd nt.05.nhr nt.07.nnd nt.09.nog nt.11.nsi nt.14.nhd
nt.00.nsq nt.03.nhi nt.05.nin nt.07.nni nt.09.nsd nt.11.nsq nt.14.nhi
nt.01.nhd nt.03.nhr nt.05.nnd nt.07.nog nt.09.nsi nt.12.nhd nt.14.nhr
nt.01.nhi nt.03.nin nt.05.nni nt.07.nsd nt.09.nsq nt.12.nhi nt.14.nin
nt.01.nhr nt.03.nnd nt.05.nog nt.07.nsi nt.10.nhd nt.12.nhr nt.14.nnd
nt.01.nin nt.03.nni nt.05.nsd nt.07.nsq nt.10.nhi nt.12.nin nt.14.nni
nt.01.nnd nt.03.nog nt.05.nsi nt.08.nhd nt.10.nhr nt.12.nnd nt.14.nog
nt.01.nni nt.03.nsd nt.05.nsq nt.08.nhi nt.10.nin nt.12.nni nt.14.nsd
nt.01.nog nt.03.nsi nt.06.nhd nt.08.nhr nt.10.nnd nt.12.nog nt.14.nsi
nt.01.nsd nt.03.nsq nt.06.nhi nt.08.nin nt.10.nni nt.12.nsd nt.14.nsq
nt.01.nsi nt.04.nhd nt.06.nhr nt.08.nnd nt.10.nog nt.12.nsi nt.nal
nt.01.nsq nt.04.nhi nt.06.nin nt.08.nni nt.10.nsd nt.12.nsq
nt.02.nhd nt.04.nhr nt.06.nnd nt.08.nog nt.10.nsi nt.13.nhd
nt.02.nhi nt.04.nin nt.06.nni nt.08.nsd nt.10.nsq nt.13.nhi

[jp.]

Oh..no. I didn't because I though there are not related with mouse.
Do I need these for mouse ?
I am downloading now, please reply if these are the one required for my analysis (I will then add them in ./ncbi-blast-2.2.28+/bin).
Thank you very much for your guidance.

Did you download all these files?

nt.00.nhd nt.02.nhr nt.04.nnd nt.06.nog nt.08.nsi nt.11.nhd nt.13.nhr
........ nt.08.nsd nt.10.nsq nt.13.nhi

[ymc]

Oh..no. I didn't because I though there are not related with mouse.
Do I need these for mouse ?
I am downloading now, please reply if these are the one required for my analysis (I will then add them in ./ncbi-blast-2.2.28+/bin).
Thank you very much for your guidance.

I am not sure but your error message said you don't have these files. Good luck!

[ymc]

Well, I finally tried soapfuse-1.26 on my tumor-normal RNA-seq data that has 100mil PE reads each.

Soapfuse ran for 7 days and crashed at step 8. At one point, the program took out 300GB of my disk space. In contrast, tophat only generated about 80GB temp files and of course it ran to completion.

Therefore, I think I have to give up on soapfuse.

I will give fusioncatcher a try later...

[ymc]

For whoever is interested, my soapfuse error at s08 is

RILP-004/503/SCARF1-001/152 encounters wrong start codon: CTG

That doesn't seem like something I know how to fix.