In Silico Design and Quality Assessment of Rep Protein Models for African Cassava Mosaic Virus

African Cassava mosaic virus (ACMV) is a begomovirus belonging to the Geminiviridae family. This single stranded DNA (ssDNA) virus severely impacts cassava production, causing significant yield losses. The Rep protein is an essential component of ACMV replication, which initiates viral DNA replication. Unfortunately, the Rep protein has not yet been crystallized, and its full three-dimensional structure remains unknown. Current computational chemistry methodologies enable the generation of reliable structural models, paving the way for rational drug design and inhibitor screening. Particularly, in this study, we compare the application of RoseTTAFold and AlphaFold in the generation of a three-dimensional model of Rep protein. Given the absence of an experimentally resolved structure, in silico validation of the generated models was performed using ERRAT, Verify3D, Whatcheck and Procheck. Results suggest that the RoseTTAFold model produced a stable model (GDT=0.78, instability index=38.13) with predominantly hydrophilic characteristics and it exhibited superior non-bounded interaction quality (ERRAT score: 95.7%).  Conversely, the AlphaFold model showed a high-confident prediction (pTM=0.5 and ipTM=0.8) featuring well-defined functional sites with slightly better stereochemical accuracy (91.8% residues in favoured Ramachandran regions). These validated models provide a robust foundation for future studies on ACMV replication and the rational design of antiviral compounds targeting the Rep protein.