Intro

Tophat was built on top of the non-splice-aware aligner Bowtie. So in order to use Tophat, you must also have Bowtie available. Tophat and Bowtie are both available as modules on Pegasus to simply load and use.

CCS does a pretty good job of keeping the list of available programs up to date with new versions installed fairly promptly, but if you wanted to, you can also install your own copy, say in order to use the same versions of all pieces of software as a collaborator or something like that. If you choose to install your own version, the CCS Student Mentors can help you do that in person, but describing that process is currently outside the scope of this tutorial. The rest of this tutorial will operate under the assumption that you are using the built-in Tophat and Bowtie modules available on Pegasus.

Setup

To load tophat and bowtie for use on pegasus, simply type the following commands:

module load bowtie2/2.2.2
module load tophat/2.0.11

These versions were the most recent available as pegasus modules at the time of writing this guide (June 29th, 2015). When new ones become available, simply use those version numbers instead. To see a complete list of all available modules, type:

module avail

Now, to confirm that the modules have been loaded properly, type:

which tophat

which should return:

/share/apps/tophat/2.0.11/tophat

similarly for bowtie2:

which bowtie2

returns:

/share/opt/bowtie2/2.2.2/bowtie2

Great, now that the software is loaded, we need to give it information about the genome to which we want to align our data. This involves three steps:

• First, we need to download the genome sequence as a fasta file
• Second, we need to download (or construct) an indexed version of the genome for bowtie2 to work with
• Third, we need to download an annotation file containing all the known genes and transcripts for this genome. This third step is technically optional, but it helps to improve the accuracy of splice junction calling and is generally recommended. We are also going to need this file when we quantify the transcripts present, so might as well use it now too.

Before we get started downloading these files, there is an important caveat to discuss. There are two major "versions" of most genomes. One is the NCBI version, the other is the Ensembl version. They are essentially the same set of As,Ts,Cs and Gs but the NCBI version starts counting the nucleotides on each chromosome from 0, while Ensembl starts counting from 1. It doesn't matter which version you use, both are perfectly fine, but you need to be consistent and get all three of the above file sets from the same source. If you mix them up, you will have a lot of confusion to deal with in any downstream analysis, especially analysis that might involve annotations of regulatory elements or SNPs.

For this guide, I am going to use Ensembl. At the time of writing, the most recent genome annotation was GRC38 release 80. So that is what we will use for each model system:

Genome sequence file:

Human: (ftp://ftp.ensembl.org/pub/release-80/fasta/homo_sapiens/dna/Homo_sapiens.GRCh38.dna.primary_assembly.fa.gz)
Mouse: (ftp://ftp.ensembl.org/pub/release-80/fasta/mus_musculus/dna/Mus_musculus.GRCm38.dna.primary_assembly.fa.gz)

I would recommend loading this Genome Sequence file onto Pegasus and putting it in the following location:
~/RNASeq/GRCh38
or
~/RNASeq/GRCm38

Find other species here: (ftp://ftp.ensembl.org/pub/release-80/fasta/)

When selecting which genome assembly file to download (for other species), select the primary_assembly instead of the top-level (if such a distinction is available for the species). The top level contains information about additional haplotypes that will not be useful in RNA-Seq analysis. You should also select the file that is called .dna. or .dna_sm. rather than .dna_rm. These phrases refer to the masking of the genome: .dna. is unmasked, .dna_sm. is 'soft-masked' by making repeat regions be lowercase nucleotides (this makes no difference to any aligners that I know of). The .dna_rm. files have repeat regions converted to Ns, but note that this will cause issues with RNA-Seq analysis by artificially removing part of your signal (not to mention that masking is far from perfect anyway).

Obviously, when new genome annotations come out, use the new release instead of release 38, version 80 that we are using now.

Genome index file:

Pre-built indexes for Bowtie2 can be found on its home page: http://bowtie-bio.sourceforge.net/bowtie2/index.shtml However, it does not have options for the newest version of the Ensembl version of the genome yet for mice or humans, so we will just have to build this index ourselves (which users of other species will probably have to do anyway). Bowtie2 has a command to build this index out of the genome sequence file. Navigate to the folder containing your genome sequence file. Let's say it is at
~/RNASeq/GRCh38/Homo_sapiens.GRCh38.dna.primary_assembly.fa.gz Then change to that directory:

cd ~/RNASeq/GRCh38

rename the fasta file to something a bit easier to deal with:

mv Home_sapiens.GRCh38.dna.primary_assembly.fa.gz GRCh38.fa.gz

Unzip it:

gunzip GRCh38.fa.gz

then enter this command:

bowtie2-build -f GRCh38.fa GRCh38

This will create a series of files called GRCh38.1.bt2, GRCh38.rev.1.bt2, GRCh38.2.bt2, etc.

Genome annotation: Human: (ftp://ftp.ensembl.org/pub/release-80/gtf/homo_sapiens/Homo_sapiens.GRCh38.80.gtf.gz)
Mouse: (ftp://ftp.ensembl.org/pub/release-80/gtf/mus_musculus/Mus_musculus.GRCm38.80.gtf.gz)
Find other species here: (ftp://ftp.ensembl.org/pub/release-80/gtf/)
Place and unzip this annotation file in the same folder as the genome sequence I would also recommend renaming the annotation file the same way we renamed the genome squence file (just for ease of use):

mv Homo_sapiens.GRCh38.80.gtf.gz GRCh38.gtf

Now we are ready to actually run Tophat!

Head over to the Running Tophat Page!