{"id":888,"date":"2023-08-15T11:16:00","date_gmt":"2023-08-15T18:16:00","guid":{"rendered":"https:\/\/labs.wsu.edu\/scads\/?page_id=888"},"modified":"2025-08-19T13:32:12","modified_gmt":"2025-08-19T20:32:12","slug":"bioinformatics","status":"publish","type":"page","link":"https:\/\/labs.wsu.edu\/scads\/research\/bioinformatics\/","title":{"rendered":"Bioinformatics"},"content":{"rendered":"\n

Bioinformatics<\/h1>\n\n\n\n
\n\n\n\n

Software Tool for Short Sequence Repeat Analysis<\/strong><\/p>\n\n\n\n

Patterns of short-sequence repeats (SSRs) at specific loci are widely used to distinguish microbial strains. In collaboration with Prof. Kelly Brayton<\/a> (Allen School for Global Animal Health, WSU), we developed RepeatAnalyzer<\/em>\u2014a robust software tool for managing and analyzing SSR data on bacterial strains infecting cattle. The initial release included an Anaplasma marginale<\/em> dataset and supported storing, searching, and identifying (i) SSR motifs, (ii) strains containing those motifs, and (iii) the publications reporting them. The prototype was deployed to track A. marginale<\/em> strains in South Africa, Asia, and multiple U.S. groups. Although our first focus was A. marginale<\/em>, the software was designed to remain flexible for any species whose strains are defined by SSRs.<\/p>\n\n\n\n

\n\n\n\n

Sequence Similarity Network Model and Applications<\/strong><\/p>\n\n\n\n

Sequence similarity networks are useful for classifying and characterizing biologically important\u202fproteins and other structures. Threshold-based approaches to similarity network construction using exact distance measures are\u202fprohibitively slow to compute and rely on the difficult task of selecting an appropriate threshold, while similarity\u202fnetworks based on approximate distance calculations compromise useful structural information. <\/p>\n\n\n\n

We introduced the Directed Weighted Average Nearest-Neighbor (DiWANN) framework that overcomes both of these drawbacks. DiWANN creates sparse but information-rich similarity graphs by representing each sequence by a node and connecting the node via a directed edge to only the closest sequence, or sequences in the case of ties. The weights on the edges represent the edit distance between the sequences. Our DiWANN model is accompanied by an efficient pruning and bounding-based algorithm for constructing the network. This loss-aware pruning lets researchers map relationships across tens of thousands of genomes without the noise or computational burden of dense networks, revealing clusters and outliers that dense graphs often obscure.<\/p>\n\n\n\n

Building on DiWANN\u2019s concise representation, we examined various applications, including analysis of cancer\u2019s driver-mutation landscape through two complementary graph views. A reduced-data DiWANN sequence-similarity network links tumors with highly similar driver-gene profiles, spotlighting cancer types whose samples cluster tightly\u2014and may therefore respond to common therapies. In parallel, a bipartite tumor-to-gene network (and its one-mode gene projection) shows which driver genes frequently co-occur and which remain exclusive to particular cancers. Together, these network perspectives uncover both pan-cancer and cancer-specific driver mutations, offering insight into tumour heterogeneity and potential therapeutic targets.<\/p>\n\n\n\n

\n\n\n\n

Papers<\/h2>\n\n\n\n