naturalis / bio-cipres Goto Github PK

View Code? Open in Web Editor NEW

3.0 12.0 4.0 24.08 MB

Phylogenomic analysis on the CIPRES REST portal

License: MIT License

Perl 97.91% Shell 2.09%

bio-cipres's Introduction

Bio::Phylo::CIPRES

Phylogenomic analysis on the CIPRES REST portal

Prerequisites

Usage of CIPRES requires a DEVELOPER account (not a normal user account) for the CIPRES REST API (CRA), and a registration for the app corvid19_phylogeny.With the account and app key, you can then populate a YAML file cipres_appinfo.yml thusly, substituting the fields with pointy brackets with the appropriate values:

---
URL: https://cipresrest.sdsc.edu/cipresrest/v1
KEY: <app key>
CRA_USER: <user>
PASSWORD: <pass>

Additional prerequisites, which should be resolved automatically during your chosen installation procedure (conda, cpanm) are listed under the PREREQ_PM field in the file Makefile.PL.

Installation

CPANM

$ cpanm Bio::Phylo::CIPRES

Example workflow

1. Aligning sequences

To align sequences in a FASTA file with MAFFT:

cipresrun \
     -t MAFFT_XSEDE \
     -p vparam.anysymbol_=1 \
     -i <infile> \
     -y cipres_appinfo.yml \
     -o output.mafft=/path/to/outfile.fasta

By adding the -v (or --verbose) flag, the XML returned by the server is shown. In the last status check, this will show additional values for -o, e.g. to retrieve STDERR and other outputs.
Most other parameters shown on the REST documentation page can also be used.
The output is written to a file with the same name is the output field (i.e. in this case a file called output.mafft), which optionally ends up in a -wd working directory.

2. Inferring trees

To infer trees from an aligned FASTA file using IQTree:

cipresrun \
    -t IQTREE_XSEDE \
    -p vparam.specify_runtype_=2 \
    -p vparam.specify_dnamodel_=HKY \
    -p vparam.bootstrap_type_=bb \
    -p vparam.use_bnni_=1 \
    -p vparam.num_bootreps_=1000 \
    -p vparam.specify_numparts_=1 \
    -i /path/to/outfile.fasta \
    -y cipres_appinfo.yml \    
    -o output.contree=/path/to/tree.dnd

bio-cipres's People

Contributors

Stargazers

Watchers

Forkers

inutano rvosa ambarishk jraspass

bio-cipres's Issues

Push container to Docker hub

Once all the individual steps (e.g. #3, #4) are working and are orchestrated using CWL (#5), an essential step for redistribution is that the Docker container does not have to be rebuilt by other users but can just be pulled down by the cwl-runner. For this to work, the container should be pushed to Docker hub. This will likely result in a tag that matches the name of this repo, i.e. naturalis/covid19-phylogeny under the docker hub naturalis organization

Write and submit manuscript

The hackathon preprint will be submitted to BioHackrxiv and will be based on the manuscript template in doc/paper.md

Rerunnable workflow as CWL

The goal of the basic workflow is to be able to consume unaligned FASTA, align this (i.e. solve #3) and build a tree with it (by addressing #4). These steps are implemented with tools, scripts, and web service calls that are all provisioned inside a Docker container (whose Dockerfile is in the root of the repo, and whose tag will be the same as the repo name).

Subsequently, these steps will be chained together using CWL, most of which is already scaffolded in PR #1. The essential test is therefore that we should be able to run the whole thing on a clean computer using something like cwl-runner. We will then submit this to covid19.workflowhub.eu.

Make bioconda package

Release the cipresrun REST client as a bioconda package

Tree construction with IQTree

IQTree is more or less the standard for building trees from viral genomes. A somewhat costly step that it does by default is to do full model testing to decide on the right model. Perhaps as per Rambaut et al. we might just settle on HKY85+G.

For the clade assignments, bootstrap support would be good (e.g. 1000 replicates), so experiment with how costly that is (-bnni parameter, I think).

Sequence alignment with MAFFT or Muscle

The experiences detailed here (nextstrain/ncov#268) show that doing the MSA in one big run eventually becomes prohibitive. This was not a problem for the 400 GenBank genomes set, but as those submissions are increasing (or when we add GISAID data) it becomes an issue.

MAFFT has the virtue of being the standard that is now being used (e.g. by Rambaut et al.) but it might be slower than Muscle (@rvosa's subjective experience)? Both can be run on the CIPRES cluster. Test and decide.

Documentation

Please specify more to add onto Documentation.md

Remove cURL dependency

i.e. rewrite requests to LWP::UserAgent