Azka Javaid

This paper introduces an autoencoder framework for denoising spatial transcriptomics networks and estimating cell signaling, showing improved performance over SVD on simulated data and biologically plausible results on real data.

SCAPE predicts cell-level cytokine activity from scRNA-seq data using gene sets and a modified VAM method, validated on simulated and real datasets.

STREAK improves receptor abundance estimation in scRNA-seq by using supervised learning and gene set scoring, outperforming existing methods on human tissue datasets.

SPECK estimates cell surface receptor abundance from scRNA-seq data using an unsupervised approach, outperforming other methods on human receptors and multiple tissues.