In Malaria journal ; h5-index 51.0
BACKGROUND : Gene Regulatory Networks (GRN) produce powerful insights into transcriptional regulation in cells. The power of GRNs has been underutilized in malaria research. The Arboreto library was incorporated into a user-friendly web-based application for malaria researchers ( http://malboost.bi.up.ac.za ). This application will assist researchers with gaining an in depth understanding of transcriptomic datasets.
METHODS : The web application for MALBoost was built in Python-Flask with Redis and Celery workers for queue submission handling, which execute the Arboreto suite algorithms. A submission of 5-50 regulators and total expression set of 5200 genes is permitted. The program runs in a point-and-click web user interface built using Bootstrap4 templates. Post-analysis submission, users are redirected to a status page with run time estimates and ultimately a download button upon completion. Result updates or failure updates will be emailed to the users.
RESULTS : A web-based application with an easy-to-use interface is presented with a use case validation of AP2-G and AP2-I. The validation set incorporates cross-referencing with ChIP-seq and transcriptome datasets. For AP2-G, 5 ChIP-seq targets were significantly enriched with seven more targets presenting with strong evidence of validated targets.
CONCLUSION : The MALBoost application provides the first tool for easy interfacing and efficiently allows gene regulatory network construction for Plasmodium. Additionally, access is provided to a pre-compiled network for use as reference framework. Validation for sexually committed ring-stage parasite targets of AP2-G, suggests the algorithm was effective in resolving "traditionally" low-level signatures even in bulk RNA datasets.
van Wyk Roelof, van Biljon Riëtte, Birkholtz Lyn-Marie
Artificial intelligence, Gene regulatory network, Machine learning, Malaria, Plasmodium falciparum