9 November 2018 in Nature Communications
Single cell messenger RNA sequencing (scRNA-seq) provides a window into transcriptional landscapes in complex tissues. The recent introduction of droplet based transcriptomics platforms has enabled the parallel screening of thousands of cells. Large-scale single cell transcriptomics is advantageous as it promises the discovery of a number of rare cell sub-populations. Existing algorithms to find rare cells scale unbearably slowly or terminate, as the sample size grows to the order of tens of thousands. We propose Finder of Rare Entities (FiRE), an algorithm that, in a matter of seconds, assigns a rareness score to every individual expression profile under study. We demonstrate how FiRE scores can help bioinformaticians focus the downstream analyses only on a fraction of expression profiles within ultra-large scRNA-seq data. When applied to a large scRNA-seq dataset of mouse brain cells, FiRE recovered a novel sub-type of the pars tuberalis lineage.
1 July 2021 in Big Data Research
2 September 2019 in arXiv
We present a new way of constructing an ensemble classifier, named the Guided Random Forest (GRAF) in the sequel. GRAF extends the idea of building oblique decision trees with localized partitioning to obtain a global partitioning. We show that global partitioning bridges the gap between decision trees and boosting algorithms. We empirically demonstrate that global partitioning reduces the generalization error bound. Results on 115 benchmark datasets show that GRAF yields comparable or better results on a majority of datasets. We also present a new way of approximating the datasets in the framework of random forests.