The use of genomic information has gained increasing relevance in epidemiology and public health, particularly the inference of genetic networks of highly evolving pathogens like HIV. This approach offers objective evidence that allows for tracking transmission dynamics, and guiding intervention. However, the virus's high genetic diversity, combined with variations in risk factors and transmission rates among affected populations, leads to substantial differences in epidemic dynamics across different epidemiological contexts. Additionally, transmission networks inferred using a distance threshold often create many spurious edges among closely related sequences, resulting in abiological transmission cycles. Management of these networks as new sequences are generated also fuel a desire to track cluster growth over time. It is therefore advisable to develop molecular analysis tools that better capture the unique landscape of each epidemic, providing a portrayal that more closely mirrors the real epidemic scenario. This dissertation introduces novel tools aimed at optimizing genetic network analysis for epidemiological purposes, with a focus on HIV-1 but with stated potential further application in other pathogens. Due to their greater convenience compared to phylogenetic methods, the inference of transmission networks through genetic distance methods has been widely used for near-real-time surveillance of HIV-1. Published research using the HIV-TRACE software package frequently uses the default threshold of 0.015 substitutions/site for HIV \textit{pol} gene sequence. In Chapter 2, Weaver \textit{et al.} introduces AUTO-TUNE, a heuristic scoring method to adaptively tune the distance threshold, to prevent giant cluster formation and maximize cluster numbers. This method was applied to various outbreaks, considering regional or temporal differences, to identify clusters with specific risk factors. For instance, we found the 0.015 threshold suits US-like epidemics, but a lower threshold of 0.005 better captures outbreaks like the CRF07 BC subtype among MSM in China. Conversely, a larger threshold is needed for regions where diverse risk factors and sparse sampling occur over longer periods. This adaptive approach aids public health officials in making more informed interventions. Genetic distance and phylogenetic tools often yield densely connected clusters, with many spurious edges that create unrealistic cycles, overestimating connectivity and inflating node degrees. In Chapter 4, we present a scalable approach, Conditionally Orthogonal Vertices For Edge Filtering in Epidemics (COVFEFE), to prune spurious edges through a straightforward application of the PC (unroll) algorithm originally developed for Bayesian graphical models. COVFEFE removes edges likely to be indirect, preserving cluster membership. It eliminates 29\%–80\% of spurious edges in simulated transmission chains with low intra-host mutation rates and 4\%–39\% in real HIV-1 epidemic data, often simplifying dense clusters to chains. Filtered networks align better with theoretical models, affecting analysis and modeling methods that use node or edge properties. COVFEFE is part of the HIV-TRACE molecular network inference package. The heterogeneity in risk factors and characteristics of the HIV-1 epidemic worldwide demands adapting bioinformatics tools for different contexts to meet local needs. In Chapter 4, we present HIV-TRACE GO, an extension of the HIV-TRACE software designed to enhance global HIV molecular epidemiology surveillance. This web application, developed primarily in JavaScript, enables real-time molecular surveillance with features tailored for multi-jurisdictional data management and internationalization. HIV-TRACE GO has been successfully implemented at CIENI-INER, a WHO-accredited HIV-1 sequencing reference laboratory in Mexico, demonstrating its feasibility for both local and international applications. We discuss the software architecture, key customizations, and its potential impact on public health surveillance. / Biology