Machine learning can differentiate venom toxins from other proteins having non-toxic physiological …

Introduction Falling costs of tandem mass spectrometry for shotgun proteomics have made generating vast amounts of protein sequence data increasingly affordable, yet the gap between obtaining these sequences and then assigning biological function to them continues to widen (Bromberg et al., 2009). Often, most sequences are deposited into protein databases with little, if any, accompanying experimental data from which biological functions can be inferred. Customarily, biological function is deduced indirectly by comparing amino acid sequence similarity to other proteins in large databases to calculate a ranking of proteins with respect to the query sequence. Using simple pair-wise comparisons as a sequence searching procedure, the BLAST suite of programs (for example, BLASTp) was first of its kind and has gained almost unprecedented acceptance among scientists (Neumann, Kumar & Shalchian-Tabrizi, 2014). Variations…


Link to Full Article: Machine learning can differentiate venom toxins from other proteins having non-toxic physiological …