Informatics & Visualisation Laboratory
The focus of informatics lab is on data analytics and visualization as well as molecular visualization and systems biology, computational systems on data storage and retrieval and is the Research Unit’s primary connection to the UCT Medical School. Using data from the Medical School, as well as techniques such as machine learning, bioinformatics and visualization analytics the Informatics Lab outputs visualized gene expressions. These gene expressions are for the purpose of cancer diagnostics. Upregulated genes are then investigated as targets for drug design by the Computational GlycoEnzymology Group.
The resources associated with the informatics lab are a) a nine-screen wall allowing for analysis of multiple data set models and UHD biomolecular structures; and b) a IBM DS3512 with two expansion drawers storage attached network (SAN).
REFERENCES:
Jahansha Ashkani, & Kevin J. Naidoo Glycosyltransferase Gene Expression Profiles Classify Cancer Types and Propose Prognostic Subtypes. Scientific Reports 6, 26451, (2016).
Christos Ferles, Yannis Papanikolaou, & Kevin J. Naidoo, Denoising Autoencoder Self-Organizing Map (DASOM). Neural Networks 105, 112-131, (2018).
Tharindu Senapathi, Simon Bray, Christopher B. Barnett, Bjoern Gruning, & Kevin J. Naidoo, Biomolecular Reaction and Interaction Dynamics Global Environment (BRIDGE). Bioinformatics 35, 3508-3509, (2019).
Christopher B. Barnett, Kiyoko Aoki-Kinoshita, & Kevin J. Naidoo, The Glycome Analytics Platform: an integrative framework for glycobioinformatics. Bioinformatics 32, 3005-3011, (2016).
-
Machine Learning and Bioinformatics Group
The main objective of the Machine Learning and Bioinformatics group at SCRU is to develop methods for Cancer and Respiratory Infection data analytics research where we process NGS produced data and perform analytics on this as well as data from publicly accessible databases. Using a systems biology approach we aim to understand the role of glycoenzymes in cause respiratory infection as well as tumourigenesis.
We employ stati
stics and bioinformatics methods such as multivariate analysis and gene expression profiling. The core focus of the research in the Bioinformatics members of the group is developing computational strategies that provide novel biological insights into human disease mechanisms and to introduce potential biomarkers for early diagnosis. The machine learning group members develop unsupervised learning methods aimed at data analytics for noisy biological data. The Denoising Autoencoder Self Organising Map (DASOM) is our signature software whereupon we develop deep learning as well as Growing hierarchical variations of this method.
The members of the Informatics group have developed an efficient Biomolecular Reaction and Interaction Dynamics Global Environment (BRIDGE) platform based on Galaxy for high through put screening. The virtual HTS of TSA inhibitor families (Hits) based on the findings from the reaction dynamics studies undertaken in the Computational GlycoEnzymology Group. These families are then narrowed down to leads using cheminformatics methods and passed on to the Glycobiomedical Laboratory for synthesis, testing and systems data development.
-
Laboratory Infrastructure and Resources
The SCRU hardware platform includes state of the art GPU clusters, data servers and infiniband clusters. The computational capability is modest and designed for code development, modellling testing and sort runs prior to using either the national HPC facility (CHPC) or the UCT and regional facility (Ilifu). Our computing enviromnent is set up for computational software developed in-house, from academic developers and commercially licenced. These have been modularized to allow users to manage their environment either in an interactive session or a batch job. Our software modules also enhance users to dynamically change environments User workstations in the computation and modeling lab perform computations on the GPU and CPU cluster compute nodes.
-
Machine Learning and Bioinformatics Group
The main objective of the Machine Learning and Bioinformatics group at SCRU is to develop methods for Cancer and Respiratory Infection data analytics research where we process NGS produced data and perform analytics on this as well as data from publicly accessible databases. Using a systems biology approach we aim to understand the role of glycoenzymes in cause respiratory infection as well as tumourigenesis.
We employ stati
stics and bioinformatics methods such as multivariate analysis and gene expression profiling. The core focus of the research in the Bioinformatics members of the group is developing computational strategies that provide novel biological insights into human disease mechanisms and to introduce potential biomarkers for early diagnosis. The machine learning group members develop unsupervised learning methods aimed at data analytics for noisy biological data. The Denoising Autoencoder Self Organising Map (DASOM) is our signature software whereupon we develop deep learning as well as Growing hierarchical variations of this method.The members of the Informatics group have developed an efficient Biomolecular Reaction and Interaction Dynamics Global Environment (BRIDGE) platform based on Galaxy for high through put screening. The virtual HTS of TSA inhibitor families (Hits) based on the findings from the reaction dynamics studies undertaken in the Computational GlycoEnzymology Group. These families are then narrowed down to leads using cheminformatics methods and passed on to the Glycobiomedical Laboratory for synthesis, testing and systems data development.
-
Laboratory Infrastructure and Resources
The SCRU hardware platform includes state of the art GPU clusters, data servers and infiniband clusters. The computational capability is modest and designed for code development, modellling testing and sort runs prior to using either the national HPC facility (CHPC) or the UCT and regional facility (Ilifu). Our computing enviromnent is set up for computational software developed in-house, from academic developers and commercially licenced. These have been modularized to allow users to manage their environment either in an interactive session or a batch job. Our software modules also enhance users to dynamically change environments User workstations in the computation and modeling lab perform computations on the GPU and CPU cluster compute nodes.
