BCC2020

→ Abstract


The presenter(s) will be available for live Q&A in this session (BCC West).

Jean-Tristan Brandenburg 1, Shaun Aron 1, Shakuntala Baichoo 2, Gerrit Botha 3,
Christopher J Fields 4, Sumir Panji 3, Nicola Mulder 3, Scott Hazelhurst 1,5

1 Sydney Brenner Institute for Molecular Bioscience, University of the Witwatersrand,
Johannesburg, South Africa.
2 University of Mauritius, Mauritius.
3 Computational Biology Division, University of Cape Town, South Africa.
4 High Performance Computing in Biology, University of Illinois at Urbana­Champaign, United States.
5 School of Electrical & Information Engineering, University of the Witwatersrand, Johannesburg, South Africa

Project Website: https://h3abionet.org/
Source Code: https://github.com/h3abionet/h3agwas,
Licence: MIT Licence

Genome Wide Association Studies (GWAS) require several complex analyses of large data sets
over many iterations. Many projects in the Human Heredity and Health in Africa (H3A)
consortium are conducting GWAS. As part of its mandate to support H3A, the Pan-African
Bioinformatics Network for H3Africa (H3ABioNet), comprising 27 research groups in 17
countries, is building scientific workflows for African scientists [Baichoo et al 2018, doi:
10.1186/s12859-018-2446-1]. Many African research groups work in resource constrained
environments with over-burdened bioinformaticists and system administrators compared to
better resourced groups elsewhere. These African research groups need access to
heterogeneous computing resources and the ability to quickly deploy consistent powerful and
reproducible workflows to different environments (cloud/local/remote with or without
containers, scheduling systems) without spending hours or days setting up the environments.

In response, we built the H3AGWAS workflows to support various GWAS phases. The pipeline is
highly scalable and portable. Built using the Nextflow language and containers (Docker and
Singularity supported) it can run in different environments (single computer, SLURM, PBS,
Amazon Batch). H3AGWAS provides powerful, flexible and parameterisable workflows
supporting quality control, data format conversion, association testing using several GWAS
tools and techniques (PLINK, GEMMA, BOLTLMM), support of genotype and imputed data, post-
association analysis and annotation, meta-analysis, heritability estimation, and fine mapping.
The key workflows provide comprehensive reports to the user.

The workflow has been used by two H3A groups (AWI-Gen & Trypanogen) and other scientists
inside and outside of Africa. Although primarily designed for production work, the flexibility
and portability of the workflows have made them useful for training.

Since its original publication, significant functionality in association testing (e.g. BOLTLMM),
post-association analysis and support for new computing environments has been added.
H3AGWAS is undergoing active development. We plan to add genotype calling from image data.
H3ABioNet is also actively building other workflows to support African science for training and
production work.

Acknowledgements: H3ABioNet is funded by a NIH Common Fund Award / NHGRI Grant
Number U24HG006941. We thank Lerato Magosi, Rob Clucas, Eugene de Beste, Aboyomini
Mosaku, Don Armstrong, Ananyo Choudhury, Ayton Meintjes, Michele Ramsay and
Dhriti Sengupta for their support.