RepinPop

Installation

Compilers, build system and dependencies.

The following packages (linux, debian based distro) are required:

• java (>=11)
• gcc (or alternativ C compiler)
• libgsl-dev
• andi
• build-essential
• cmake (>=3.11)
• phyloml

Install dependencies on debian based linux distros (debian, *ubuntu, mint, ...)

sudo apt install linux-libc-dev util-linux git make gcc build-essential libgsl-dev gsl-bin andi wget zip unzip phyml

Create the conda environment

$> conda env create -n repinpop --file=environment.yml

This step will create a new conda environment named repinpop and install a number of dependencies from the conda repository. For details, inspect the file environment.yml which lists the dependencies and version requirements.

Activate the new environment:

$> conda activate repinpop

Build and install cmake targets.

RAREFAN uses cmake to configure, build, and install most of its non-python dependencies including the RAREFAN java code andclustDist Cluster Distances into Phylogenies for phylogenetic analysis. The installation target directory is the $CONDA_PREFIX directory.

We observed that clustDist produces faulty results if compiled inside the conda environment. As a workaround, we recommend to build clustDist with deactivated conda environment. Nevertheless, we install clustDist and other build products into the conda environment.

Record the value of the $CONDA_PREFIX environment variable, e.g.

echo $CONDA_PREFIX > conda_prefix.txt

Deactivate the conda environment.

conda deactivate

Build in separate build directory.

$> mkdir build
$> cd build
$> cmake -DCMAKE_INSTALL_PREFIX=$(cat ../conda_prefix.txt) ..

The last line instructs sets the installation prefix to $CONDA_PREFIX.

 $> make

This fetch, build, and install the dependencies into the conda environment created in the first step.

Set library path.

Some environment variables (in particular LD_LIBRARY_PATH) have to be set explicitly.

source setenv.sh

Commandline interface (CLI)

We provide a standalone commandline interface to run rarefan. See the [Releases]{https://github.com/mpievolbio-scicomp/rarefan/releases} section of this repository.

The command line signature is as follows:

$> rarefan [-h] [-o OUTDIR] -r REFERENCE [-c MIN_NMER_OCCURRENCE] [-l NMER_LENGTH] [-d DISTANCE_GROUP_SEEDS] -q QUERY_RAYT [-e E_VALUE_CUTOFF] [-R] [-j THREADS] [-t TREEFILE] [-i] DIR

  DIR                   Contains the genome DNA sequences and RAYT AA sequences to be analyzed.

optional arguments:
  -h, --help            show this help message and exit
  -o OUTDIR, --outdir OUTDIR
                        Results will be written to OUTDIR. OUTDIR will be created if not existing (default: ./rarefan_out).
  -r REFERENCE, --reference REFERENCE
                        Filename of the reference genome sequence
  -c MIN_NMER_OCCURRENCE, --min_nmer_occurrence MIN_NMER_OCCURRENCE
                        Only Nmers of NMER_LENGTH that occur more frequently than MIN_NMER_OCCURRENCE will be taken into account (default: 55). See RAREFAN manual for details.
  -l NMER_LENGTH, --min_nmer_length NMER_LENGTH
                        Only Nmers of NMER_LENGTH that occur more frequently than MIN_NMER_OCCURRENCE will be taken into account (default: 21). See RAREFAN manual for details.
  -d DISTANCE_GROUP_SEEDS, --distance_group_seeds DISTANCE_GROUP_SEEDS
                        Set the distance between the REPIN groups (???)
  -q QUERY_RAYT, --query_rayt QUERY_RAYT
                        Filename or path of the amino acid sequence file containing the RAYT protein sequence (default: None).
  -e E_VALUE_CUTOFF, --e_value_cutoff E_VALUE_CUTOFF
                        e-value cutoff for tblastn of the query rayt sequence against the submitted genomes (default: 1e-30).
  -R, --no-repins       Do not analyse REPINS (default: False).
  -j THREADS, --num_threads THREADS
                        Number of threads for parallel cluster analysis with MCL (default: 4).
  -t TREEFILE, --treefile TREEFILE
                        Filename or path of the phylogenetic tree of submitted genomes (newik format, '.nwk' extension). If none given and more than four genomes are submitted, the tree will be calculated and written to
                        OUTDIR/tmptree.nwk (default: tmptree.nwk).
  -i, --interactive     Interactive mode. Ask for confirmation before starting the analysis run.

Running the RAREFAN web server

Database backend

The webserver uses MongoDB as a backend. Install mongodb-server, create a database user named 'rarefan', secured by password, and a database 'rarefan'. Assign the 'dbAdmin' role for the database 'rarefan' to the 'rarefan' user. Consult the mongodb manual if unsure how to do this.

Configuration

Copy the configuration template app/config_template.py to app/config.py and edit the settings. An example is given below.

import os

class Config(object):
    SECRET_KEY = 'supersecretkey'
    SERVER_NAME = 'localhost:5000'
    MONGODB_SETTINGS = {
        'db': 'rarefan',
        'host': 'localhost',
        'port': 27017,
        'username': 'rarefan',
        'password': 'RaReF@npw01'
    }
    REDIS_URL = os.environ.get("REDIS_URL") or 'redis://'
    MAIL_SERVER = 'mail.my.server.com'
    MAIL_USERNAME='rarefan@mail.my.server.com'
    MAIL_PASSWORD='mailpass'

    MAIL_USE_TLS=True
    MAIL_USE_SSL=False
    MAIL_PORT=25

    MAIL_DEBUG=False
    DEFAULT_MAIL_SENDER='rarefan@mail.my.server.com'

Redis Job Queue

Jobs submitted to RAREFAN are processed by redis. In your conda environmont, install rq and redis.

$> conda install rq redis

Starting the server

To launch the server, run

$> flask run 

And navigate your browser to http://localhost:5000 .

R shiny app

RAREFAN output can be visualized using our provided R scripts and shiny app. The mongo db password must be set in shinyapp/analysis/server.R in order to query the run parameters from the database.

To start the shiny app, run

$> Rscript run_app.R

from the project root directory.

NOTE: Data visualisation on a local deployment server is currently not working.

Testing

The directory test/scripts/ contains two scripts:

dl_zenodo.sh

dl_zenodo.sh may be used to download reference datasets from zenodo, and unpack the data into the directory test/data/datasets/. Datasets can be downloaded individually or together.

Syntax:

./dl_zenodo.sh [all | neisseria | chlororaphis | dokdonia] 

test.sh

Having downloaded the reference datasets (see above), the functionality of RAREFAN can be tested using the script test.sh.

Syntax:

./test.sh DATASET OPERATION1 [OPERATION2 [OPERATION3 [...] ] ]

DATASET is the name of one of the reference datasets (neisseria, chlororaphis, or dokdonia). Further test datasets are contained in test/data and may also be used, e.g. neisseria_small which only contains a small subset of the reference neisseria dataset.

OPERATION is at least one of:

• clean: Remove all data from a previous run
• setup: Copy reference data to /tmp/rarefan_test/DATASET and configure the environment for running RAREFAN.
• run_java: Run the java code REPINecology to compute RAYT and REPIN populations from the input data.
• run_andi: Calculate distances between the given genomes.
• run_clustdist: Generate a phylogenetic tree based on output from andi.
• plots: Generate figures from the output data (Phylogenetic trees for input genomes and RAYTs, REPIN and RAYT population sizes, and correlation between replication rate and REPIN population size.)
• ref_plots: Generate figures from the reference output data that comes with the downloaded dataset (originally in DATASET/out/, copied to /tmp/rarefan_test/DATASET/ref in the setup operation).
• test_java: Checks if a file results.txt was produced.
• test_andi: Checks if the file .dist file was generated by andi.
• test_clustdist: Checks if the .nwk treefile was generated by clustDist.
• test_plots: Checks if all plots were generated.
• test_ref_plots: Checks if all plots were generated from the reference output data.
• test_md5: Computes md5 checksums for all datafiles in the output directory (except subdirectories) and compares to checksums in the directory test/md5/.

Docker

We provide a docker container that packs all dependencies of the java backengine (java code).

Pull the container

To pull the most recent docker container, run (in a terminal)

docker pull mpievolbioscicomp/repinpop_base

Our gitlab-CI configuration is setup to use this docker image as a basis of the build procedure.

End-to-end testing in docker

Scripts in CI/ are used for testing and building. You can also utilize these scripts for local testing:

docker run --mount source=rarefan_testing,target=/mnt/rarefan_test mpievolbioscicomp/repinpop_base /e2e.sh /mnt/rarefan_test/docker_test.log 

This command will launch the docker container, build the RAREFAN code and run a test suite. The entire log output will be written to the file /mnt/rarefan_test/docker_test.log on the container. Note that /mnt/rarefan_test is setup as a persistent docker mount point, i.e. it can be visited after the test run is complete and the container shut down. The easiest is to navigate to /var/lib/docker/volumes/rarefan_testing/_data on the host machine.