Lectures and seminars MEB faculty event: Johanna Simin and Ina Schuppe Koistinen from The Centre for Translational Microbiome Research (CTMR)

Challenges in microbiome data analysis

Ina Schuppe Koistinen, Fredrik Boulund, Johanna Simin, Lars Engstrand

The Centre for Translational Microbiome Research (CTMR) builds on a deep understanding of translational microbiome research and has established a broad technical, biological, clinical and epidemiological platform for studying complex microbiological communities in well-defined human materials. CTMR forms a solid foundation for understanding the contribution of the microbiome to normal physiology and pathophysiology and opens opportunities for development of novel therapies in gastroenterology and women’s health. CTMR’s research in women’s health aims to describe the microbiome in healthy women of reproductive age as well as in pregnant women. Large clinical cohorts have been collected to investigates associations between the vaginal, fecal and oral microbiota and the risk for diseases, such as HPV infection, pregnancy complications, recurrent pregnancy loss and endometriosis. In addition, CTMR also works with large nationwide health and administration data registries, with a primary focus on studying the long-term effects of prescription drugs on the risk of cancer and Clostridioides difficile infection, recurrence, and mortality. In the women’s health area, the association of prescription drugs with multiple outcomes among mothers and children are studied.

Metagenomic microbiome data consists of millions of observations of fragments of the constituent microbial genomes, with varying forms of noise applied to the measurements in the form of incorrect nucleotide base calls, uneven representations of the underlying distribution of genomic material, etc. Often, the observations are represented as relative abundances computed from observed feature counts, which are inherently sparse and compositional. The interpretation of metagenomic data is therefore non-trivial and dependent on dedicated computational and statistical methods that can handle the unique properties of the data. While the microbiome research community has put efforts into developing statistical frameworks and effective tools to handle these challenges, work still remains to develop and establish robust statistical methods to analyze microbiome data collected according to common study design patterns. 

Our presentation will highlight the challenges related to the complexity of the multidimensional data generated from large clinical cohorts and modelling of drug-microbiome interactions. We wish to explore possibilities for a future collaboration with MEB researchers.

Zoom link: https://ki-se.zoom.us/j/67119392703