search and evaluate gene orthologs using hmm and phylogeny

ortholotree was written to first and foremost look for gene orthologs. Its functionality would suggest that it can do more than this, but this is not its intent by design. It can be run directly through CLI, and the options are hopefully explained sufficiently in the --help option to allow relatively straightforward usage. More complete documents may follow in time, if there is enough demand...

To install, change into the downloaded directory and type:

python setup.py install

You may need sudo privileges depending on your system.

This will install a config file that can be automatically generated by running config.py, or by manually populating it with file paths. The format is simply two columns: KEY PATH The necessary keys are determined by the modules, and therefore cannot be arbitrarily set. However, if you initialise the file with config.py, it will put all the necessary KEYs in the file, even if it cannot find the appropriate file path.

orthotree relies on the *hmmer (for protein searching - http://hmmer.janelia.org) *mafft (for sequence alignment - http://mafft.cbrc.jp/alignment/software/) *RAxML (for phylogeny construction - http://sco.h-its.org/exelixis/web/software/raxml)

you will also need to install the following python libraries: *Biopython *Matplotlib *Numpy

ortholotree will work best with a couple of iterations of usage, depending on your starting point.

1st pass

If you wish to find the ortholog of a particular gene, then you will start by providing the peptide ID to the -g flag, or the link to a fasta file containing a single sequence using the -f flag. This will run the phmmer search on all protein databases using the query peptide. It will then filter the results according to three parameters:

*Local threshold (-t) For each species searched, collect all matches that are above this value. If the species of the query sequence is provided, and a peptide database exists for this species, then the best match is the score of the query matched to itself. All other scores are therefore relative to the best score. If the species is either unknown or not in a db, then the best score is the highest score returned for each peptide db searched.

*Minimum collection (-c) For each species, the best n results will be collected, regardless of whether they pass the local threshold. If you only want genes that pass the local threshold, then pass a value of 0 to this flag.

*Global threshold (-a) Once all genes passing the previous two filters are collated, only genes with a score that is above the specified percentage of the best score will be kept.

2nd pass

Using the results of the first pass, select likely ortholog candidates (this may simply be the gene in your target species, as well as your initial query gene). Pass all genes as a comma-separated list to the -g flag, and specify -B to tell the program to build a hidden markov model using an alignment of the specified proteins.

The various peptide databases will then be searched using the hmm, providing a more sensitive search. You can filter the hits using the same parameter flags as in the first pass, but typically you can use higher percentage values (as the orthologs will generally be a better match to the model than to a single peptide, particularly from highly divergent species.

BE WARNED! If you provide non-orthologous genes to the model, you will probably pull out multiple gene families, giving too many genes, or not enough.

3rd pass

This final pass will often not be necessary, but if you want to tidy up your results, then use the tree built by the 2nd pass to identify the orthologs from all species available, and use all identified orthologs to build the hmm model as specified above. You can also use some additional flags to help remove unwanted hits:

*Exclude gene (-x) a comma-separated list of genes that you do not want in your alignment or phylogeny.

*Exclude species (-e) like the -x flag, only all genes from a species will be excluded. Species names are specified using the four-letter key used to construct the peptide database paths (E.g. Cerapachys biroi is Cbir).

*Maximum length (-l) to prevent gene fusions from being picked up and causing havoc with your alignment and phylogeny, you can specify a length (in amino acids), for which all proteins must be smaller than to remain.

If you would like more information on how robust your phylogeny is, use the -b flag to run a bootstrap analysis. This will produce 100 bootstrapped trees, then show the bootstrap support on the (independently calculated) most likely tree.