Page 81 - University of Pretoria Research Review 2017
P. 81

         Foreword
Introductory Messages
DEVELOPMENT AND CHANGE
Into the Future
Bioinformatics empowering modern biological science
Fourie Joubert, Centre for Bioinformatics and Computational Biology
PEOPLE AND CONTEXTS
HEALTH AND WELL-BEING
PLANET AND SUSTAINABILITY
Awards
Lead Researchers
79
  The nature of experimental approaches to biological research has changed rapidly in recent years, particularly with the advent of omics- based technologies. These include genomics, transcriptomics, proteomics, metabolomics and other related fields.
These technologies often generate massive datasets for many molecules in parallel, and are regularly applied to hundreds or even thousands of samples simultaneously. The raw outputs are massive datasets, swiftly expanding the big data pool for biological science. These datasets require extensive pre-processing before analysis, often increasing the amount of data by a factor of x10 to generate analysis- ready data. High-powered statistical approaches are
then required for these datasets to generate useful knowledge.
While classical statistical analyses
have long been crucial for analysing
large datasets, the wealth of
information now becoming
available is fast necessitating new
approaches to the detection of intricate patterns in such datasets, and the application of this knowledge to new data. Machine learning approaches and deep learning tactics are steadily becoming the standard
to mine such large biological datasets. The models developed in this way can subsequently be applied
to newly-generated biological data to rapidly extract information from it. This is valuable for research performed in developing countries, where knowledge gained from massive experimental studies costing millions, can be extrapolated for use in smaller affordable local studies. Local population-specific genetic diversity, however, may often necessitate the adaptation of models built with data from populations in developed regions of the world.
A first step to gauge the suitability of models built for large European and Asian populations to South African ethnic groups was the South African Human Genome Project, which was published in Nature Communications in 2017 (see page 76). Professor
Fourie Joubert at the Centre for Bioinformatics
and Computational Biology was involved in identifying variants in the genomes of Sotho, Xhosa
and Coloured individuals to highlight differences to other population groups where big datasets are available. Another project to highlight differences
in genes involved with breast cancer in African vs European populations is currently
ongoing in the Centre. Importantly, the H3Africa (Human Heredity and
Health in Africa) consortium has recently designed and produced genotyping resources customised to African populations. The widespread application of these resources in African research studies will greatly contribute to knowledge of South African and African genetic diversity.
 




































































   79   80   81   82   83