EP 64: Machine Learning and Mutli-Omics Data with Mxolisi Nene

In this episode of the Bioinformatics Lab Podcast, Mxolisi Nene shares his journey from a curious kid “scanning soil” with a stick and a broken Pentium II in rural KwaZulu-Natal to a bioinformatician and PhD candidate at the Agricultural Research Council in Pretoria. He walks through his path from animal science into bioinformatics, profiling the gut microbiomes of indigenous village chickens using 16S and metagenomic sequencing, and how wrestling with messy real-world data led him into multi-omics integration and machine learning. Mxolisi explains concepts like feature engineering, neural networks, and ecological “tipping points” in soil ecosystems—showing how combining metagenomic, metabolomic, proteomic, and genomic layers can help predict when an environment is on the brink of collapse, with implications for agriculture, food security, and even disease research.
We also dig into the philosophical side of his work: why the explosion of public omics data makes it almost a moral obligation to use these tools for better outbreak prevention and environmental stewardship, how conferences like PHA4GE in Cape Town and the AI working group are quietly seeding a new generation of multi-omics scientists, and what it feels like to realize that the five-year-old kid obsessed with dirt grew up to do exactly what he was pretending to do—only now with HPC clusters, neural nets, and GitHub.