Phylomizer is an automated phylogenetics pipeline which aims to reproduce the three main step follow by traditional phylogeneticists. These steps, as shown in the image, are 1) Homology search, 2) Multiple Sequence Alignment Reconstruction, and 3) Phylogenetic Tree reconstruction.
Phylomizer can be used to reconstruct a single gene tree. However, it aims to assist on the reconstruction of large collection of phylogenies. In fact, Phylomizer has been used in phylomedb.org to generate more than 5.6 million phylogenetic trees, as January, 2016.
There are two scripts included in the Phylomizer package. Please be ware that by default the multiple sequence alignment used for phylogenetic tree reconstruction is generated as follow: 1) three different alignment programs are used to generate alignments in forward and reverse sequences orientation; 2) these 6 alignments are combined into a consensus one using M-Coffee; 3) final alignment is generated by trimAl after removing those columns from the consensus alignments with a low consistency score e.g. no column has been seed in any of the 6 alignment used to generate the consensus one.
phylomizer.py allows the user to perform the whole phylogenetic pipeline, specific steps e.g. homology searches, or specific combinations e.g. homology searches + multiple sequence alignment reconstruction
## Just reconstructing the best fitting evoluionary model for a ML tree.
$> ./source/phylomizer.py -i input_msa_alignment --steps trees -c config_file -o output_folderHere we have access to all available options which are shared (most of them) by both scripts.
## Check input options for phylomizer.py - assuming we are at the ROOT directory of the package
$> ./source/phylomizer.py -h
optional arguments:
-h, --help show this help message and exit
-i INFILE, --in INFILE
Input file containing the query sequence/s
--min_seqs MINSEQS Set the minimum sequences number to reconstruct an alignment/tree.
This parameter overwrites whatever is set on the config file.
--max_hits MAXHITS Set the maximum accepted homology hits after filtering for e-value/coverage.
This parameter overwrites whatever is set on the config file.
--steps [{all,homology,alignments,trees} [{all,homology,alignments,trees} ...]]
Set which step/s should be performed by the script
-d DBFILE, --db DBFILE
Input file containing the target sequence database
--cds CDSFILE Input file containing CDS corresponding to input protein seqs
-c CONFIGFILE, --config CONFIGFILE
Input configuration file
-o OUTFOLDER, --out OUTFOLDER
Output folder where all generated files will be dumped
-p PREFIX, --prefix PREFIX
Set the prefix for all output files generated by the pipeline
-r, --replace Over-write any previously generated file
--version show program\'s version number and exit
-v {0,1,2,logfile,none,stderr}, --verbose {0,1,2,logfile,none,stderr}
Set how information should be dumped. It could be used levels or tags
It overwrites whatever is set on the configuration file.pipeline.py aims to perform the three main steps implemented in the pipeline following the configuration set on the configuration file.
## Given a single input sequence, retrieve its homologous sequences from a predefined DB
## using e.g. BLAST, reconstruct the alignment for this set of sequences, and then the
## best fitting ML tree.
$> ./source/pipeline.py -i input_single_sequence_file -d input_DB_file -c config_file -o output_folderOne of the main aim of this pipeline is to facilitate the generation of thousands of single gene trees. Therefore it is possible to take a given sequences file and generate an appropriate data structure to store all this data. Moreover, the auxiliary script prepare_data.py will assist you in the process.
A couple of things should be taken into account. First, the species tag is intended to follow the PhylomeDB species code e.g. HUMAN which are automatically derived from sequence IDs e.g. Phy00086SJ_HUMAN > HUMAN. Alternatively, species code can be extracted from the first 3 letter of any sequence ID e.g. Sce000078 > Sce. If not species tag is given, the script will process all sequences in the input file.
Second, a ROOT project folder will be generated. It will contains three main folders: 1) Data, where all single gene trees will be stored. To avoid file storage problems, gene trees are divided into subfolders of a given size e.g. 1000. Subfolders size can be changed using this parameter --size; 2) BlastDB where all databases are stored. Remember that input databases should be formatted prior to run the pipeline using e.g. formatdb; 3) jobs will contain a file with a call to the pipeline for each of the gene trees to be reconstructed. This file could be easily used in the context of High Performance Computing facilities and their queue system. This folder will also contain the configuration file in order to allow specific configurations for each project.
## prepare_data.py could be as easy as below
$> ./source/prepare_data.py --folder ROOT_FOLDER --seed_sp HUMAN --script 'PATH TO source/pipeline.py'
-c 'PATH TO config/config.pipeline --db 'PATH to input sequences file'Configuration File example. This file includes all currently supported programs and the standard parameters used with all of them. Additional parameters and execution modes e.g. MPI call, can be set editing this file.
[PhylomeDB Phylomizer Configuration file KMM](config/PhylomeDB Configuration - KMM). It uses 3 different aligner (KAlign, Muscle, and Mafft) to generate a consensus alignment of 3 aligners x 2 sequences orientation using M-Coffee. Later on this consensus alignment is processed by trimAl to remove low consistency columns. The use of KAlign instead of DiAlign-TX considerably speeds up the whole alignment generation process with a sligthly degratation on performance. Therefore, this configuration is the default one in PhylomeDB, as January 2016.
[PhylomeDB Phylomizer Configuration file DMM](config/PhylomeDB Configuration - KMM). It uses 3 different aligner (DiAlign-TX, Muscle, and Mafft) to generate a consensus alignment of 3 aligners x 2 sequences orientation using M-Coffee. Later on this consensus alignment is processed by trimAl to remove low consistency columns.