Star with cufflinks

[Nino]

Hey Guys,

I was wondering if anyone knows how to integrate cufflinks with star. Up until now I have star just for the alignments with chimeric searches, now I need to use cufflinks since searches for novel transcripts. I was told the two programs are used hand in hand. If anyone has any insight on this please let me know. Tried to search information regarding this but was unsuccessful.

Thanks,
Nino

[alexdobin]

Hi Nino,

For non-strand-specific data, you need to use STAR option --outSAMstrandField intronMotif which will add the XS attribute to all canonically spliced alignments using their introns' motifs - that's exactly what Cufflinks needs.

For strand-specific data, you do not need any extra parameters for STAR runs, but you need to use --library-type option for Cufflinks. For example, for the "standard" dUTP protocol you need to use --library-type fr-firststrand in Cufflinks.

Do not use the option --outSAMattributes All, as Cufflinks (as well as many other tools) have compatibility problems with the latest SAM format specifications.

For poly-A+ selected RNAs, all other default parameters generally work well in our ENCODE experience.
As always I would highly recommend using annotations for mapping.
Also, --outFilterType BySJout could help clean up rare junctions, and --outFilterIntronMotifs RemoveNoncanonicalUnannotated will clean up non-canonical junctions.

Total RNA data is much harder to assemble with Cufflinks since it contains a substantial intronic signal. Please check this post, you may need to mask "complex" loci.

Cheers
Alex

[Nino]

Hey Alexdobin,

Thanks for the info and help I was able to get it to work. Cheers!

Nino

[bob-loblaw]

Hi guys,

this is probably a good place to ask this question. But I ran STAR with that option to add the XS attribute. But when I run cufflinks I get this

BAM record error: found spliced alignment without XS attribute
BAM record error: found spliced alignment without XS attribute
BAM record error: found spliced alignment without XS attribute
BAM record error: found spliced alignment without XS attribute
BAM record error: found spliced alignment without XS attribute
> Processing Locus chr1:2835641-2836005 [ ] 0%BAM record error: found spliced alignment without XS attribute
> Processing Locus chr1:2882539-2882944 [ ] 0%BAM record error: found spliced alignment without XS attribute
> Processing Locus chr1:3536416-3537049 [ ] 0%BAM record error: found spliced alignment without XS attribute
> Processing Locus chr1:3537448-3539171 [ ] 0%BAM record error: found spliced alignment without XS attribute
> Processing Locus chr1:3646678-3647038 [ ] 0%BAM record error: found spliced alignment without XS attribute

It does keep running, and the percentage does increase, but I was wondering if this behaviour is normal for STAR and cufflinks?

[sdriscoll]

No, not expected. These errors are saying that the XS tag is not added. I've used this workflow as well and didn't have any trouble. Maybe STAR slipped up.

[bob-loblaw]

I've had issues with this version of STAR on my machine before. The option for writing out unmapped reads to fastq format doesn't work either, although theres a pretty easy workaround for that. I may try to reinstall it. Thanks.

Good thing STAR is so quick to run! If it was Tophat that did this, I'd have to set aside about a month of machine time to catch up!

[alexdobin]

Please try the latest STAR patch ftp://ftp2.cshl.edu/gingeraslab/trac...TAR_2.3.1o.tgz
It supposed to have fixed the problems with the output of unmapped fastqs.
I also checked that it outputs the XS flag. If it still does not work, please send me your Log.out file.

[bob-loblaw]

Please try the latest STAR patch ftp://ftp2.cshl.edu/gingeraslab/trac...TAR_2.3.1o.tgz
It supposed to have fixed the problems with the output of unmapped fastqs.
I also checked that it outputs the XS flag. If it still does not work, please send me your Log.out file.

The unmapped fastx is working for me now as is cufflinks, it started working with the version I had installed though. Odd. Although I've noticed with the intronmotif option for cufflinks that it only works for me when it's very last option specified.

Thanks

[alexdobin]

The unmapped fastx is working for me now as is cufflinks, it started working with the version I had installed though. Odd. Although I've noticed with the intronmotif option for cufflinks that it only works for me when it's very last option specified.

Thanks

This is indeed odd. Could you please send me the Log.out file for the case where the intronMotif option does not work? It actually contains information about all parameters that STAR read from command line.

[bob-loblaw]

This is indeed odd. Could you please send me the Log.out file for the case where the intronMotif option does not work? It actually contains information about all parameters that STAR read from command line.

Here is the log file from one where the introMotif option does not work! Thanks!

[alexdobin]

Here is the log file from one where the introMotif option does not work! Thanks!

Hi @bob-loblaw,

According to the Log.out, I believe there was a typo in your command line:
--outSAMstrandField introMotif
while you needed
--outSAMstrandField intronMotif
Unfortunately STAR does not yet check all of the input values so in this case the parameter just got ignored and the XS flag was not generated.

Cheers
Alex

[bob-loblaw]

Hi @bob-loblaw,

According to the Log.out, I believe there was a typo in your command line:
--outSAMstrandField introMotif
while you needed
--outSAMstrandField intronMotif
Unfortunately STAR does not yet check all of the input values so in this case the parameter just got ignored and the XS flag was not generated.

Cheers
Alex

Well I've never felt as stupid as I do right now haha. Thanks though!

[alexdobin]

I have made these kind of typo mistakes many times, and sometimes even had to run a debugger to figure out what the problem was.
It is high on my TODO list (though not on the very top) to enforce strict control of all user options.

[ndejay]

Hi all,

I had an additional question.

You say that --outSAMstrandField intronMotif is necessary for unstranded reads, but what would happen if I used the option on stranded reads? I am hoping it has no effect.

Thanks.

[sdriscoll]

I'd assume it would just do what it does normally except in the output the intron orientation would always match the alignment orientation. The intron motif and orientation is estimated from the genome and not from the orientation of the alignment. There should be a setting in STAR, however, to make it aware of this so it can correctly assign intron orientation to novel intron motifs...if there are any.