paveethranswam,github

analysing-single-cell-rna-and-atac-sequence-data

▪ Integrated real-world single cell RNA and ATAC sequence data using unsupervised Manifold-learning based maximum mean discrepancy (MMD) measure to jointly embed RNA and ATAC sequences. ▪ Evaluated data integration using FOSCTTM score – and applied hyper-parameter tuning to yield an average FOSCTTM score close to zero, for better alignment performance. ▪ Performed dimensionality reduction by comparing Non-Matrix Factorization and Latent Dirichlet Allocation methods on genome sequences. ▪ Clustered the data after dimensionality reduction using K-Means and used UMAP to visually check the clustering results, and finally evaluated if the separate RNA and ATAC are co-assayed and compared cluster overlap fraction using DICE Coefficient values. ▪ Generated a graph to represent physical protein-protein interactions and performed Graph partitioning by comparing min-cut and normalized-cut algorithms. ▪ Future work would include evaluating the partition performance by using Gene Ontology enrichment analysis to check number of enriched clusters out of 50 output clusters.

applied-algorithms

bioinformatics-assignment-1

bioinformatics-lab-github-demo

cuml

cuML - RAPIDS Machine Learning Library

data-science-books-1

diet-nutrition-and-foods-recommendation

To deliver high level research in two directions, named statistical analysis of molecular omics data and nature language processing analysis of literature data. Need to develop new mathematical models and computational approaches to integrate the findings from the two data sources. The expected deliverables include (1) new computational model, (2) inference of highly explainable dependencies among biological or biomedical features.