Tophat building Bowtie index from gtf file

[Aholton]

Not really a problem, as much as an inquiry. When I run Tophat it tells me that it is building a Bowtie index from my .gtf file I've supplied. However, I already supplied a Bowtie2 index in the command.

Was wondering if I am doing something wrong or if it needs both types of indexes?

[Aholton]

Ok just checked and my CuffDiff output file gene_exp.diff is showing the first three columns as the same thing instead of showing me the geneid and gene so I'm assuming something is definitely wrong in my process.

I got the 22 chromosome files (including x and y) from NCBI and the genes.gtf file from them as well. I built a Bowtie2 index with the 22 chr. files with x and y. I checked to make sure they had the same first column identifiers.

I ran Tophat for the 2 samples (and their replicates) and then took their accepted_hits.bam and used CuffDiff to find the difference. I tried many different ways but can't get the gene_exp.diff file to have information in the first three columns.

tldr; Ran Tophat then CuffDiff, gene_exp.diff file's first 3 columns are the same, not supposed to be like this,help

[Aholton]

Bump, still having the reoccuring problem.

[kmcarr]

Not really a problem, as much as an inquiry. When I run Tophat it tells me that it is building a Bowtie index from my .gtf file I've supplied. However, I already supplied a Bowtie2 index in the command.

Was wondering if I am doing something wrong or if it needs both types of indexes?

If you supply TopHat with a GTF annotation file it will first extract transcript sequences from your genome based on the annotations and build a Bowtie index of the transcript fasta file. I assume the index you supplied to TopHat is for the genome sequence. When using -G/--GTF TopHat will first attempt to align reads directly to transcripts then unaligned reads to the genome.

If you are planning on aligning several data sets to the same genome/annotation it would be a waste of time for TopHat to rebuild the transcript index every time. For this reason TopHat also has the --transcriptome-index option which you supply the first time you run TopHat, along with the -G option to direct TopHat where to store the index it builds. In subsequent runs you can omit the -G option and use the --transcriptome-index parameter to direct TopHat to where it can locate the prebuilt transcript indexes. Check out the "Supplying your own transcript annotation data:" section in the TopHat Manual.

[kmcarr]

Ok just checked and my CuffDiff output file gene_exp.diff is showing the first three columns as the same thing instead of showing me the geneid and gene so I'm assuming something is definitely wrong in my process.

I got the 22 chromosome files (including x and y) from NCBI and the genes.gtf file from them as well. I built a Bowtie2 index with the 22 chr. files with x and y. I checked to make sure they had the same first column identifiers.

I ran Tophat for the 2 samples (and their replicates) and then took their accepted_hits.bam and used CuffDiff to find the difference. I tried many different ways but can't get the gene_exp.diff file to have information in the first three columns.

tldr; Ran Tophat then CuffDiff, gene_exp.diff file's first 3 columns are the same, not supposed to be like this,help

This is expected depending on the annotation supplied.

First of all there is a slight error in the cuffdiff manual describing the format of the gene_exp.diff file; it says that there are 13 columns with the first three being Tested id, gene, locus. In the current output there are actually 14 columns with a 'gene_id' column added between test_id and gene. Here is the header and two lines from a recent output of mine (and realize that the headers don't line up directly over the corresponding data columns due to text formatting):

test_id	gene_id	gene	locus	sample_1	sample_2	status	value_1	value_2	log2(fold_change)	test_stat	p_value	q_value	significant
AT1G01080	AT1G01080	AT1G01080	1:45295-47019	fae1_7-8	fae1_9-10	OK	23.3894	15.1937	-0.622382	1.24939	0.211521	0.706688	no
AT1G01090	AT1G01090	PDH-E1 ALPHA	1:47484-49286	fae1_7-8	fae1_9-10	OK	609.513	569.592	-0.0977292	0.153759	0.8778	0.999999	no

You can see that in the first line the test_id, gene_id and gene are all the same, whereas in the second there is a common name in the gene column. This is all a function of what information is present in your annotation (GTF) file which cufflinks/cuffdiff is able to parse out.

[Aholton]

Ok thank you so much! That makes much more sense now