Tuning Stochastic Dynamic Models of the Dynamics of DNA Replication in Trypanosomatids

• Ano de defesa: 2025
• Autor(a) principal: Scholl, Bruno Boaventura
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Computational Model; DNA; Modelo Computacional; Replicação; Replication; Transcrição; Transcription; Tripanossoma; Trypanosoma
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/95/95131/tde-29062025-202725/

The Chagas disease and the Human African trypanosomiasis are classified as neglected tropical diseases (NTDs), which afflict 1 billion people worldwide and are more prevalent in poorer countries. These two diseases are caused by the Trypanosoma cruzi and Trypanosoma brucei, respectively, protozoan endoparasites of the Trypanosomatida family. Trypanosomatids change their form depending on their lifecycle, host, and infection phase by changing their surface proteins, called variant surface glycoproteins (VSGs). Their genes are organized in polycistronic regions, long sections of multiple genes with a single promoter, and are constantly transcribed, even during DNA replication in the S phase of the cell cycle. Interactions between replication and transcription are increased, and head-on collisions can interrupt replication and cause lesions and mutations in the DNA. These interactions and collisions could define the landscape of origin firing during the S phase, and computational model simulations can help us better understand these interactions by providing a large volume of data from large sample sizes. In this work, we present the tuning of the ReDyMo, a model of the DNA replication and interactions between replication and transcription, for two specific organisms, Trypanosoma brucei and Trypanosoma cruzi, so that they are predictive of the time of replication of the bases of the genome. For the Trypanosoma cruzi models, we tuned two models: one using MFA-Seq to derive the probabilities of activation of origins in the genome and one using ChIP-Seq data with the specific locations of constitutive origins. The models were tuned for more than 1000 trials each and achieved SMAPE errors of 7.93% for Trypanosoma brucei, 5.55% for the Trypanosoma cruzi with MFA-Seq, and 8.39% for the Trypanosoma cruzi with ChIP-Seq. Analyzing the tuning of these models, we concluded that although they are predictive in some sections of the genome, in other parts, they diverge from the expected replication time. Tuning them with more trials and different tuning settings could yield more accurate models of the DNA replication in Trypanosomatids. We conducted a preliminary analysis of the model&#8217s output characteristics and found no patterns of head-on collisions in the edges of polycistronic regions, where sections with no transcription meet sections with active transcription. Moreover, we identified a pattern of increased head-on collisions in sections of the genome that replicate early in the S phase. This pattern is present in the Trypanosoma brucei and the Trypanosoma cruzi models with MFA-Seq. The models provide some level of accuracy and can be improved through the tools and tuning framework developed in this work. These models can help researchers better understand the dynamics of DNA replication in trypanosomatids and investigate the possible links between these replication-transcription interactions and mutations and the parasites&#8217 ability to escape the hosts immune system.