Best source for GTF file for use with TopHat/Cufflinks

sdarko · 07-13-2011, 04:48 AM

I've been grabbing the "refSeq genes" table (human, hg19) from UCSC in GTF file format for use with TopHat/Cufflinks.

I was just curious as to what everyone else is using and if I might find a more optimal GTF file to use.

gringer · 07-13-2011, 06:06 AM

That's what I've been using, because I can understand how to use the UCSC table browser to output a GTF file (just change the output format). Just be mindful that it updates fairly frequently, and you might discover new annotations by looking again at something you mapped a few months ago.

gavin.oliver · 07-14-2011, 03:42 AM

Ensembl provides GTF files with each build as standard and is much more comprehensive than Refseq alone. I have switched to using the genome and GTF from Ensembl as a result.

sdarko · 07-14-2011, 06:03 AM

Quote:

Originally Posted by gavin.oliver

Ensembl provides GTF files with each build as standard and is much more comprehensive than Refseq alone. I have switched to using the genome and GTF from Ensembl as a result.

Thanks for the advise. I will try it out today.

sdarko · 07-19-2011, 06:59 AM

Quote:

Originally Posted by gavin.oliver

Ensembl provides GTF files with each build as standard and is much more comprehensive than Refseq alone. I have switched to using the genome and GTF from Ensembl as a result.

Did you get your genome and gtf from the ensembl website?

Other than renaming the gtf and fasta header entries so that they matched, was there anything else that you had to do to make the gtf file work with tophat? I'm getting an error that my gtf doesn't contain junctions.

gavin.oliver · 07-19-2011, 07:16 AM

I didn't have to do anything else, no.

What command are you using to execute Tophat?

sdarko · 07-19-2011, 07:32 AM

Quote:

Originally Posted by gavin.oliver

I didn't have to do anything else, no.

What command are you using to execute Tophat?

I want to make sure that I'm not messing up anything too basic first.

I grabbed the reference genome in fasta format from here --> ftp://ftp.ensembl.org/pub/release-63...o_sapiens/dna/

I grabbed the associated GTF file from here -->ftp://ftp.ensembl.org/pub/release-63/gtf/homo_sapiens/

I then process them so that the entry names match in both files.

Here are a few lines from my GTF from ensembl:

Code:

chr18           protein_coding  exon    49501   49557   .       -       .        gene_id "ENSG00000173213"; transcript_id "ENST00000308911"; exon_number "1"; gene_name "RP11-683L23.1"; transcript_name "RP11-683L23.1-001";
chr18           protein_coding  CDS     49501   49557   .       -       0        gene_id "ENSG00000173213"; transcript_id "ENST00000308911"; exon_number "1"; gene_name "RP11-683L23.1"; transcript_name "RP11-683L23.1-001"; protein_id "ENSP00000309431";
chr18           protein_coding  start_codon     49555   49557   .       -       0        gene_id "ENSG00000173213"; transcript_id "ENST00000308911"; exon_number "1"; gene_name "RP11-683L23.1"; transcript_name "RP11-683L23.1-001";
chr18           protein_coding  exon    49129   49237   .       -       .        gene_id "ENSG00000173213"; transcript_id "ENST00000308911"; exon_number "2"; gene_name "RP11-683L23.1"; transcript_name "RP11-683L23.1-001";
chr18           protein_coding  CDS     49129   49237   .       -       0        gene_id "ENSG00000173213"; transcript_id "ENST00000308911"; exon_number "2"; gene_name "RP11-683L23.1"; transcript_name "RP11-683L23.1-001"; protein_id "ENSP00000309431";
chr18           protein_coding  exon    48940   49050   .       -       .        gene_id "ENSG00000173213"; transcript_id "ENST00000308911"; exon_number "3"; gene_name "RP11-683L23.1"; transcript_name "RP11-683L23.1-001";
chr18           protein_coding  CDS     48940   49050   .       -       2        gene_id "ENSG00000173213"; transcript_id "ENST00000308911"; exon_number "3"; gene_name "RP11-683L23.1"; transcript_name "RP11-683L23.1-001"; protein_id "ENSP00000309431";
chr18           protein_coding  exon    47390   48447   .       -       .        gene_id "ENSG00000173213"; transcript_id "ENST00000308911"; exon_number "4"; gene_name "RP11-683L23.1"; transcript_name "RP11-683L23.1-001";
chr18           protein_coding  CDS     47393   48447   .       -       2        gene_id "ENSG00000173213"; transcript_id "ENST00000308911"; exon_number "4"; gene_name "RP11-683L23.1"; transcript_name "RP11-683L23.1-001"; protein_id "ENSP00000309431";
chr18           protein_coding  stop_codon      47390   47392   .       -       0        gene_id "ENSG00000173213"; transcript_id "ENST00000308911"; exon_number "4"; gene_name "RP11-683L23.1"; transcript_name "RP11-683L23.1-001";
chr18           miRNA   exon    48162   48272   .       +       .        gene_id "ENSG00000221441"; transcript_id "ENST00000408514"; exon_number "1"; gene_name "AP001005.1"; transcript_name "AP001005.1-201";
chr18           protein_coding  exon    158483  158714  .       +       .        gene_id "ENSG00000101557"; transcript_id "ENST00000261601"; exon_number "1"; gene_name "USP14"; transcript_name "USP14-201";
chr18           protein_coding  CDS     158699  158714  .       +       0        gene_id "ENSG00000101557"; transcript_id "ENST00000261601"; exon_number "1"; gene_name "USP14"; transcript_name "USP14-201"; protein_id "ENSP00000261601";
chr18           protein_coding  start_codon     158699  158701  .       +       0        gene_id "ENSG00000101557"; transcript_id "ENST00000261601"; exon_number "1"; gene_name "USP14"; transcript_name "USP14-201";
chr18           protein_coding  exon    163308  163453  .       +       .        gene_id "ENSG00000101557"; transcript_id "ENST00000261601"; exon_number "2"; gene_name "USP14"; transcript_name "USP14-201";
chr18           protein_coding  CDS     163308  163453  .       +       2        gene_id "ENSG00000101557"; transcript_id "ENST00000261601"; exon_number "2"; gene_name "USP14"; transcript_name "USP14-201"; protein_id "ENSP00000261601";
chr18           protein_coding  exon    166787  166819  .       +       .        gene_id "ENSG00000101557"; transcript_id "ENST00000261601"; exon_number "3"; gene_name "USP14"; transcript_name "USP14-201";
chr18           protein_coding  CDS     166787  166819  .       +       0        gene_id "ENSG00000101557"; transcript_id "ENST00000261601"; exon_number "3"; gene_name "USP14"; transcript_name "USP14-201"; protein_id "ENSP00000261601";

Here are the names of my chromosomes according to bowtie-inspect:

Code:

chr1
chr2
chr3
chr4
chr5
chr6
chr7
chr8
chr9
chr10
chr11
chr12
chr13
chr14
chr15
chr16
chr17
chr18
chr19
chr20
chr21
chr22
chrX
chrY
chrM

In TopHat, I get the following error:

Code:

[Mon Jul 18 15:15:21 2011] Reading known junctions from GTF file
        Warning: TopHat did not find any junctions in GTF file

In Cufflinks, I get the following error:

Code:

[08:34:37] Loading reference annotation.
Error: duplicate GFF ID 'ENST00000445581' encountered!

gavin.oliver · 07-19-2011, 07:38 AM

The only difference I can see between my setup and yours is that I removed the 'chr' prefixes. There was a reason for this - but I can't remember what it was!

gringer · 07-19-2011, 07:43 AM

Well, if chromosome 22 is anything to go by, it looks like the chromosome labels in the fasta file don't include the 'chr' bit.

sdarko · 07-19-2011, 08:09 AM

Well, I figured out the problem. I put an extra tab between the chrom name and the second column in the gtf file when I renamed everything.

I fixed it and all is well.

gavin.oliver · 07-19-2011, 08:10 AM

Quote:

Originally Posted by sdarko

Well, I figured out the problem. I put an extra tab between the chrom name and the second column in the gtf file when I renamed everything.

I fixed it and all is well.

Glad to hear it

hbt · 10-18-2011, 03:07 AM

@sdarko Could you explain little how you went about "renaming the gtf and fasta header entries so that they matched" please.
I'm keen to update the gtf I use with tophat to the ensembl version.

many thanks for any advice you may be able to give

shurjo · 10-24-2011, 09:46 AM

Look here: http://cufflinks.cbcb.umd.edu/igenomes.html

kopi-o · 10-24-2011, 11:18 AM

If you download both the genome FASTA and the annotation from ENSEMBL, you shouldn't need to rename anything.

HSV-1 · 10-11-2012, 08:57 PM

Hi, gavin,
I have analysed my RNA-seq data with the references (both genome reference and annotation reference) from UCSC and ensemble. What I found is that the map results with the reference from UCSC is much more than those with the reference from ensemble. What I don't understand is that how this is possible?
What is confusing me much more is that GTF from UCSC is less than half of the one in Ensemble. With a smaller reference I got more results !
Do you have any idea?
Thanks!

Quote:

Originally Posted by gavin.oliver

Ensembl provides GTF files with each build as standard and is much more comprehensive than Refseq alone. I have switched to using the genome and GTF from Ensembl as a result.

carmeyeii · 12-14-2012, 12:36 PM

So you can supply TopHat with a GTF file of annotated transcripts, which, using the --GTF option, will be the first place where reads are mapped, followed by the whole genome, with or without novel junction discovery in this second stage. As I understand it, this is after TopHat 1.4.
I'm curious to know how t was before 1.4. I think you could already give TopHat a GTF file, but it used it second. Am I right? If so, what is the difference between using it [the GTF file] first and using it second after the genome?

Carmen

gringer · 12-14-2012, 12:42 PM

Quote:

I'm curious to know how it was before 1.4

I'm not sure about version numbers, but tophat's previous approach didn't try mapping to a transcriptome first. The current approach assembles a transcriptome from the GTF file and maps to that first, then does the novel junction discovery after that on the substantially smaller remainder of reads.

The previous approach was slower (and possibly less accurate), because the junction discovery would be done on all reads, rather than just the ones that didn't map to the known transcriptome.

carmeyeii · 12-14-2012, 12:48 PM

I see, but it could still use a GTF file at some point, right?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
cufflinks : analysis comparison with and without a gtf reference file	sohnic	Bioinformatics	3	07-07-2019 06:40 AM
how to get a gtf file for cufflinks	camelbbs	Bioinformatics	9	07-07-2019 06:30 AM
cufflinks won't read my GTF file	moriah	Bioinformatics	0	08-28-2011 01:31 AM
Tophat/Cufflinks GTF (canine/non-human)	M&M	RNA Sequencing	0	08-01-2011 04:00 PM
Cufflinks GTF file	ECHo	Bioinformatics	0	02-15-2010 03:59 AM

07-13-2011, 04:48 AM	#1
sdarko Member Location: Bethesda, MD Join Date: Apr 2009 Posts: 51	Best source for GTF file for use with TopHat/Cufflinks I've been grabbing the "refSeq genes" table (human, hg19) from UCSC in GTF file format for use with TopHat/Cufflinks. I was just curious as to what everyone else is using and if I might find a more optimal GTF file to use.

07-13-2011, 06:06 AM	#2
gringer David Eccles (gringer) Location: Wellington, New Zealand Join Date: May 2011 Posts: 838	That's what I've been using, because I can understand how to use the UCSC table browser to output a GTF file (just change the output format). Just be mindful that it updates fairly frequently, and you might discover new annotations by looking again at something you mapped a few months ago. Last edited by gringer; 07-13-2011 at 06:07 AM. Reason: got the URL wrong

10-24-2011, 11:18 AM	#14
kopi-o Senior Member Location: Stockholm, Sweden Join Date: Feb 2008 Posts: 319	If you download both the genome FASTA and the annotation from ENSEMBL, you shouldn't need to rename anything. Last edited by kopi-o; 10-24-2011 at 11:31 AM. Reason: clarity

07-14-2011, 03:42 AM	#3
gavin.oliver Senior Member Location: uk Join Date: Jan 2010 Posts: 110	Ensembl provides GTF files with each build as standard and is much more comprehensive than Refseq alone. I have switched to using the genome and GTF from Ensembl as a result.

07-19-2011, 07:16 AM	#6
gavin.oliver Senior Member Location: uk Join Date: Jan 2010 Posts: 110	I didn't have to do anything else, no. What command are you using to execute Tophat?

07-19-2011, 07:38 AM	#8
gavin.oliver Senior Member Location: uk Join Date: Jan 2010 Posts: 110	The only difference I can see between my setup and yours is that I removed the 'chr' prefixes. There was a reason for this - but I can't remember what it was!

07-19-2011, 07:43 AM	#9
gringer David Eccles (gringer) Location: Wellington, New Zealand Join Date: May 2011 Posts: 838	Well, if chromosome 22 is anything to go by, it looks like the chromosome labels in the fasta file don't include the 'chr' bit.

07-19-2011, 08:09 AM	#10
sdarko Member Location: Bethesda, MD Join Date: Apr 2009 Posts: 51	Well, I figured out the problem. I put an extra tab between the chrom name and the second column in the gtf file when I renamed everything. I fixed it and all is well.

10-18-2011, 03:07 AM	#12
hbt Member Location: UK Join Date: Jan 2011 Posts: 20	@sdarko Could you explain little how you went about "renaming the gtf and fasta header entries so that they matched" please. I'm keen to update the gtf I use with tophat to the ensembl version. many thanks for any advice you may be able to give

10-24-2011, 09:46 AM	#13
shurjo Senior Member Location: Rockville, MD Join Date: Jan 2009 Posts: 126	Look here: http://cufflinks.cbcb.umd.edu/igenomes.html

12-14-2012, 12:36 PM	#16
carmeyeii Senior Member Location: Mexico Join Date: Mar 2011 Posts: 137	So you can supply TopHat with a GTF file of annotated transcripts, which, using the --GTF option, will be the first place where reads are mapped, followed by the whole genome, with or without novel junction discovery in this second stage. As I understand it, this is after TopHat 1.4. I'm curious to know how t was before 1.4. I think you could already give TopHat a GTF file, but it used it second. Am I right? If so, what is the difference between using it [the GTF file] first and using it second after the genome? Carmen

12-14-2012, 12:48 PM	#18
carmeyeii Senior Member Location: Mexico Join Date: Mar 2011 Posts: 137	I see, but it could still use a GTF file at some point, right?