The Cancer Noise Atlas (TCNA): A Webserver of Multiscale Transcriptional Noise in Cancer

Cancer cells are characterized by highly divergent gene expression, resulting in expression heterogeneity in tumors, which contributes to variable therapeutic responses. The advent of next-generation sequencing has enabled the elucidation of genome-wide transcriptional variation ("noise") across tumor types. However, the current lack of data analysis pipelines for evaluating gene expression variation, their web-availability, and associated databases that catalog this variation across cancer sites, limits investigations into tumor heterogeneity. Here, we report an analytical platform for investigating transcriptional variation profiles in cancer - The Cancer Noise Atlas (TCNA), version 1.0. TCNA offers analysis and customized visualizations of transcriptional noise from bulk RNA-seq data, sourced from Genomics Data Commons (GDC), at three scales: (i) Genes, (ii) Pathways, and (iii) Tumors. The platform integrates and pipelines algorithms for analyzing transcriptional noise through tumor heterogeneity metrics such as coefficient of variation (CV), standard deviation (SD), mean absolute deviation (MAD), Differential Expression Phenotype Targeted Heterogeneity (DEPTH), and DEPTH2. Taken together, TCNA provides a significant advancement upon existing work in that it provides a simultaneous and comprehensive assessment of multiple integrated noise metrics for GDC datasets. Furthermore, with an increasing number of scRNA-seq datasets, TCNA is envisioned to grow into a high-resolution web portal for multiscale noise analysis across cancer types. Cataloging and mapping this heterogeneity can potentially assist in understanding tumor characteristics as well as disease progression and therapeutic response. The portal is freely available at https://tcna.lums.edu.pk/.