You have to create the indices yourself using the "bowtie2-build" command and the fasta genome file as argument. This step is done only once and the location of the files are passed as the argument in the tophat command.

Here's more info on bowtie-build: http://bowtie-bio.sourceforge.net/bowtie2/manual.shtml#the-bowtie2-build-indexer

Tophat is a very old method for mapping your transcripts to the transcriptome. It was superseded by tophat2 then hisat and then hisat2 some time ago. Therefore, I would recommend hisat2 https://ccb.jhu.edu/software/hisat2/index.shtml.

However, even hisat2 is now looking very old in comparison to more modern pseudoaligners such as kallisto (https://pachterlab.github.io/kallisto/about) or sailfish (https://sailfish.readthedocs.io/en/master/sailfish.html). There are many advantages to taking this approach for RNA-seq, look at the documentation and paper for each (For example, its more accurate and much faster!) .

Hi Benjamin,

As Terezinha Souza pointed out, you need to create the bowtie2 index for the genome first. You do not need to export $BOWTIE2_INDEXES. The transcriptome index you tried to create with the command, requires the genome bowtie2 index. Likewise for the alignment, you still need the genome bowtie2 index

First, start by removing the transcriptome data folder,

rm -rf transcriptome_data

Then:

bowtie2-build Tcas.fa Tcas

# this will create Tcas*bt2 in the current directory

# now create the transcriptome index

tophat2 -G Tcas.gff --transcriptome-index=transcriptome_data/Tcas Tcas

#now run tophat2 on reads test.fastq

tophat2 --transcriptome-index=transcriptome_data/Tcas Tcas test.fastq

Thank you to Terezinha Souza and Gen Lin for the thorough replies. I did figure this out eventually, but I hope this shows up on a web or researchgate search for everyone. The manual is a bit vague and misleading in this section, though I sympathize that it is difficult to delineate every single step involved. I don't doubt others will have a similar difficulty. Adam P Cribbs I am looking into Hisat2, but I am a little wary of Kallisto because of the biased estimate log-fold change estimate B it spits out. The authors say its accurate for sign (less/more expressed), but not as reliable for magnitude of change, though maybe that has changed?

Gen Lin I am beginner to RNA seq data analysis, I need to know ,To run tophat2 on reads, do I need to use the output files of trimmomatic? As trimmomatic generates four files output_forward_paired.fq.gz output_forward_unpaired.fq.gz output_reverse_paired.fq.gz output_reverse_unpaired.fq.gz

so which read should i use?

Can you please guide me. Thank you in advance