Murat Can Çobanoğlu

Haisu: Hierarchical Supervised Nonlinear Dimensionality Reduction

Haisu extends nonlinear dimensionality reduction methods to incorporate hierarchical labels, improving visualization of datasets with parent-child relationships and semi-supervised settings.

TReNCo: Topologically associating domain (TAD) aware regulatory network construction

TReNCo constructs context-specific transcription factor-gene regulatory networks using TAD boundaries, epigenetic marks, and gene expression data to model cardiac development in mice.

Fast and interpretable scRNA-seq data analysis

This paper introduces generative Bayesian Dirichlet-multinomial models for cell type annotation in scRNA-seq, showing effective transfer from bulk to single-cell data and meaningful clustering without density assumptions.

Accurate differential analysis of transcription factor activity from gene expression

EPEE improves TF activity inference by integrating regulatory networks and accounting for shared targets, showing accuracy in immune and cancer contexts.