Improving extraction of protein — Protein interaction datasets from KUPS using hashing approach

The machine learning approaches frequently address the extraction of training datasets from the online databases to build computational or mathematical models. The training data downloaded from the online server and databases are most often carry redundancy and noise. Heuristics methods are most common to filter the data. Dataset filtering process is time consuming and researcher has to do this tedious work. We propose a more generic filter to detect frequent exceptions to increase the quality of generated datasets based on Perl Hash Programming and regular expression methodology. Future development of noise and error reduction approaches is important to make use of the full potential of available database knowledge. We make use of the datasets of protein - protein interaction generated by The University of Kansas Proteomics Service (KUPS).