The broad applicability of RNA-seq data applications to different organisms with a ref- erence genome makes it the most needed across diverse research fields. This dissertation focuses on the development and enhancement of mathematical frameworks to improve the performance of evolutionary algorithms. It also provides a benchmark accuracy of HLA- callers. We first investigate the enhancement of the genetic algorithms with saltations to improve the solution quality of the NP-hard problems based on the theory of punctuated equilibrium. Saltations are induced to overcome the limitations of traditional local search heuristics, which often converge prematurely on sub-optimal solutions. By mathematically modeling these jumps, we enable the algorithm to escape local optima and navigate fitness landscapes more effectively. We validate this framework with benchmark optimization prob- lems, demonstrating a significant improvement in solution quality and algorithmic stability over standard GA implementations. Complementing these algorithmic advancements, we present a rigorous benchmarking study of alignment-based HLA callers using six datasets. Beyond the accuracy of the tools, we conduct a granular performance comparison across various HLA genes to identify specific loci that are consistently miscalled by current meth- ods. Furthermore, we examine the impact of sequencing parameters and genomic reference versions on caller, while thoroughly assessing the computational resources required for each software. Together, these contributions provide a multifaceted approach to refine the math- ematical and computational tools essential for decoding the complex molecular machinery of the cell.