#Preparing evidence data for annotation

This exercise is meant to get you acquainted with the type of data you would normally encounter in an annotation project. You will get an idea of where to download protein sequences, and also try out some programs that often are used. We will for all exercises use data for the fruit fly, Drosophila melanogaster, as that is one of the currently best annotated organisms and there is plenty of high quality data available.

##1. Obtaining data

**Swissprot:** Uniprot is an excellent source for high quality protein sequences. The main site can be found at http://www.uniprot.org. This is also the place to find Swissprot, a collection of manually curated non-redundant proteins that cover a wide range of organisms while still being manageable in size.

Exercise 1 - Swissprot:
Navigate the Uniprot site to find the download location for Swissprot in fasta-format. You do not need to download the file, just find it. In what way does Swissprot differ from Uniref (another excellent source of proteins, also available at the same site)?

**Uniprot:** Even with Swissprot available, you also often want to include protein sequences from organisms closely related to your study organism. An approach we often use is to concatenate Swissprot with a few protein fasta-files from closely related organisms and use this in our annotation pipeline.

Exercise 2 - Uniprot:
Use Uniprot to find (not download) all protein sequences for all the complete genomes in the family Drosophilidae. How many complete genomes in Drosophilidae do you find?

**Refseq:** Refseq is another good place to find non-redundant protein sequences to use in your project. The sequences are to some extent sorted by organismal group, but only to very large and inclusive groups. The best way to download large datasets from refseq is using their ftp-server at ftp://ftp.ncbi.nlm.nih.gov/refseq/.

Exercise 3 - Refseq:
Navigate the Refseq ftp site to find the invertebrate collection of protein sequences. You do not need to download the sequences, just find them. The files are mixed with other types of data, which files include the protein sequences?

**Ensembl:** The European Ensembl project makes data available for a number of genome projects, in particular vertebrate animals, through their excellent webinterface. This is a good place to find annotations for model organisms as well as download protein sequences and other types of data. They also supply the Biomart interface, which is excellent if you want to download data for a specific region, a specific gene, or create easily parsable file with gene names etc.

Exercise 4 - Ensembl Biomart:
Go to Biomart at http://www.ensembl.org/biomart/martview and use it to download all protein sequences for chromosome 4 in Drosophila melanogaster. Once you have downloaded the file, use some command line magic to figure out how many sequences are included in the file. Please ask the teachers if you are having problems here.

##2. Running an ab initio gene finder

**Setup:** For this exercise you need to be logged in to Uppmax. Follow the UPPMAX login instructions.

Before going into the exercises below, you should create in your home folder a specific folder for this practical session and add a symbolic link to a folder with the course data using:

cd ~/
mkdir practical1
cd practical1
ln -s /proj/g2015008/course_data

When you are done, you should have a folder called course_data in your practical1 folder. This course_data folder is write-proteced, it is only a resource for you to obtain data from, but not where you are writing your own outputs to!
NOTE! We do not supply full paths in all of the exercises below. You will need to find the files yourself, which will be easy since you are an expert Linux-hacker. :)

Also, we have made a genome browser called Webapollo available for you on the address http://bils-web.imbim.uu.se/drosophila_melanogaster
This browser already has a number of tracks preloaded for you, but you can also load data you have generated yourself using the ‘file” menu and then ‘open’ and ‘local files’. First time you go there you need to log in using your first name as user name and your last name as password.

**Ab initio gene finders:** These methods have been around for a very long time, and there are many different programs to try. We will in this exercise focus on the gene finder Augustus. These gene finders use likelihoods to find the most likely genes in the genome. They are aware of start and stop codons and splice sites, and will only try to predict genes that follow these rules. The most important factor here is that the gene finder needs to be trained on the organism you are running the program on, otherwise the probabilities for introns, exons, etc. will not be correct. Luckily, these training files are available for Drosophila.

Exercise 5 - Augustus: First load the needed modules using:
module load bioinfo-tools
module load augustus

Run Augustus on your genome file using:
augustus –species=fly course_data/dmel/chromosome_4/chromosome/4.fa > augustus_drosophila.gtf

Take a look at the result file using ‘less augustus_drosophila.gff’. What kinds of features have been annotated? Does it tell you anything about UTRs?

The gff-format of Augustus is non-standard (looks like gtf) so to view it in a genome browser you need to convert it. You can do this using genometools which is available on Uppmax.

Do this to convert your Augustus-file:

Module load genometools
gt gtf_to_gff3 augustus_drosophila.gtf > augustus_drosophila.gff3

Transfer the augustus_drosophila.gff3 to your computer using scp:
scp login@milou.uppmax.uu.se:/home/login/practical1/augustus_drosophila.gff3 .

Load the file in Webapollo, also display the track ‘EnsEMBLprotein’ by selecting it from the table on the left. Here find the WebApollo instruction
How does the Augustus annotation compare with the Ensembl annotation? Are they identical?

Exercise 6 - Augustus with yeast models:
Run augustus on the same genome file but using settings for yeast instead (change species to Saccharomyces).

Load this result file into Webapollo and compare with your earlier results. Can you based on this draw any conclusions about how a typical yeast gene differs from a typical Drosophila gene?

##3. Assembling transcripts based on rna-seq data

Rna-seq data is in general very useful in annotation projects as the data usually comes from the actual organism you study and thus avoids the danger of introducing errors caused by differences in gene structure between your study organism and other species.

The program Cufflinks can be used to assemble transcripts from mapped rna-seq reads. First the reads need to be mapped to the genome, and we prefer using the mapper Tophat2 as it belongs to the same family of programs as Cufflinks and is splice-aware. The result from Tophat2 is a BAM-file, a binary file with the coordinates of all mapped reads. We have in this practical already created such a file for you for chromosome 4 of D. melanogaster, and you can find it in course_data/dmel/chromosome_4/bam/.

Exercise 7 - Cufflinks:
Load the Cufflinks module using ‘module load cufflinks/2.1.1’. By typing ‘cufflinks’ you will get a list of the parameters you can change and also see the default values for each parameter.

Then run Cufflinks on the supplied BAM-file using:
cufflinks -o outdir -p 8 -b 4.fa -u accepted_hits.chr4.bam

When done you can find your results in the directory ‘outdir’. The file transcripts.gtf includes your assembled transcripts. As Webapollo doesn’t like the gtf format file you should convert it in gff3 format (cf. Exercise 5). Then, transfer the gff3 file to your computer and load it into Webapollo. How well does it compare with your Augustus results? Looking at your results, are you happy with the default values of Cufflinks (which we used in this exercise) or is there something you would like to change?

##4. Checking the gene space of your assembly.

Cegma is a program that includes sequences of 248 core proteins. These proteins are conserved and should be present in all eukaryotes. Cegma will try to align these proteins to your genomic sequence and report to you the number of proteins that are successfully aligned. This percentage can be used as a measure of how complete your assembly is.
Note: In a real-world scenario, this step should come first and foremost. Indeed, if the result is under your expectation you might be required to enhance your assembly before to go further. As this task is taking a while, we have chosen to do it at the end of this practical session. The result should be available after the lunch.

Exercise 8 - Cegma:
Here you will try Cegma on Chromosome 4 of Drosophila melanogaster.First, load cegma by typing ‘module load cegma’. The problem is that the file ‘4.fa’ has fasta-headers that are only numbers, and Cegma won’t accept that. Can you figure out how to change the fasta header to ‘chr4’ rather than just ‘4’ using the linux command sed? Ask the teachers if you are having problems, or cheat by using the already parsed file 4_parsed.fa. :)

cegma -g 4.fa -T 8

When done, check the output.completeness_report. How many proteins are reported as complete? Does this sound reasonable?