Quest for Orthologs (QfO) Benchmark pipeline with Nextflow and Docker. This branch of the repository contains the development implementation for running the QfO benchmarking http://orthology.benchmark-service.org on the OpenEBench platform. The codebase is by no means stable nor bug free.
The workflow takes as input the ortholog predictions of a method in either tab-delimited format that lists pairs of orthologs or an orthoxml file. The predictions must be done on the QfO reference proteomes, either on the dataset of 2011 or 2018. For how to retrieve the reference proteomes, please consider the instructions on the current benchmark service. The workflow will (i) validate the input predictions, (ii) convert the predictions into an internal format, and (iii) compute the benchmark metrics for various benchmarks.
Orthology predictions must be provided by the user. An example file is available in the example directory. Reference datasets have to be made available to the workflow and can be obtained online for the publicly accessible datasets. See Usage section for more details how to obtain them.
- You must have a running installation of Docker and Nextflow.
- Clone this repository
- ~~Create the necessary docker images by running
./build_dockers.sh latest~~ This is no longer needed, as the nextflow.config contains the docker tags in the container specification and will download the relevant image from dockerhub automatically.- Download the reference data by running
./fetch_reference_data.py. for the desired year. This will download the reference datasets for the store them in reference_data/<year>/. The nextflow workflow can then mount the data into the docker container.- Run the pipeline with
nextflow run main.nf -profile docker
this will launch the pipeline with the default parameters that are specified in the nextflow config file. Output files will be created by default into out/. Use nextflow run main.nf --help to obtain a list of possible parameters.