Tokenvizz is a newly developed tool designed for genomic data discovery and visualization, leveraging Graph Retrieval-Augmented Generation (GraphRAG) inspired tokenization and graph-based modeling. It employs innovative tokenization methods, including single nucleotide tokenization and byte-pair encoding (BPE) from the DNABERT2 model, alongside non-overlapping k-mer tokenization from the Nucleotide Transformer. In Tokenvizz, genomic sequences are represented as graphs, with sequence k-mers serving as nodes and attention scores as edge weights, facilitating the visual interpretation of complex, non-linear relationships within DNA sequences. Additionally, research utilizing AlphaFold2 has expanded to predict structural conformations of peptides, using a dataset of 557 peptides, marking a significant advancement in peptide analysis. AlphaFold3 has also been evaluated for its accuracy in predicting ligand-bound G protein-coupled receptor (GPCR) structures, showing improvements in overall architecture accuracy.