Supporting clinical research with analysis of large complex high dimensional datasets
The Translational Bioinformatics platform supports analysis of large complex high dimensional datasets routinely collected in studies including multi-‘omics and clinical data types which are paramount for disease stratification and subtype discovery. The platform has expertise in next-generation sequence and transcriptomics analysis, pathway and network analysis, machine learning and data integration. It also provides training to assist in engaging biomedical and clinical user communities; these focus on commonly used methods such as network and pathway analyses as well as emerging methods such as multi-omics integration for stratification.
The platform utilises the High-Performance Computer Cluster infrastructure, Rosalind, across King’s College London and in partnership with the NIHR Maudsley Biomedical Research Centre which provides a powerful combination of the raw computational power of traditional HPC computing and the flexibility and diversity of a virtual machine cloud infrastructure.
- RNAseq was used to analyse and contexualise transcriptomics in Blaschko-Linear Psoriasis (BLP) and showed that BLP has both shared and distinct features with Psoriasis Vulgaris (Onoufriadis et al, J Invest Dermatol. 2021 Jul 24).
- Multi-type data integration supported the development of models using Cervicovaginal microbiota and metabolome data to predict preterm birth risk in an ethnically diverse cohort (Flaviani et al, JCI Insight. 2021 Aug).
- Machine learning using SARS-COV-2 RNAemia and proteomic trajectories over time, was used to identify molecular features that inform prognostication in covid-19 patients (Guttmann et al., Nat Commun Jun 2021).
- The platform supported data structuring and the development of scripts for data analysis of the large data set for the study of genetic risk factors and the impact on the development of adult psychiatric disease after preterm birth (Cullen, H. et al., Sci Rep 2021, 11(1): 11443).
- Stratification of very pre-term birth children into risk subtypes, data integration and the application of network-based data fusion methods of clinical data, socio-emotional data and executive function data is being used for disease subtype discovery in very pre-term birth children (Ongoing research).