Computational workflow for predicting transcriptional modulation to achieve therapeutically desired cellular conversion for regenerative medicine

Abstract

AbstractIntroductionRegenerative medicine promises cure for currently incurable diseases and pathological conditions. Its central idea is to leverage healthy cells to regenerate diseased cells, tissues or organs through the process of cellular conversion. The most common method to achieve this is by modulating the activity of specific transcription factors. However, the large number of protein-coding genes and transcription factors in humans and their complex interactions poses a challenge in identifying the most suitable ones for modulation. Here, we propose a computational workflow that facilitates the prediction of such transcription factors for achieving desired cellular conversion, along with highlighting their mechanistic basis in terms of the gene regulatory network of the target cell type.MethodsThe proposed workflow leverages three existing computational tools: TransSynW, PAGA and SIGNET, and uses single-cell transcriptome data of the starting and target cell types as inputs.ResultsWe used this workflow on a sample dataset of human hindbrain neuroepithelial stem cells and midbrain medial floorplate progenitor cells to generate hypothesis for converting the former to the latter cell type. The workflow predicted the transcription factors for modulation and provided insight into their differential expression dynamics and influence on the predicted gene regulatory network of the target cells.ConclusionOur computational workflow helps extract meaningful predictive and mechanistic insights from high-dimensional biological data, which otherwise is difficult to accomplish from individual tools alone. We believe this workflow can help researchers generate mechanistically founded hypotheses for achieving desired cellular conversions as a step towards regenerative medicine.