PDBMine: A Reformulation of the Protein Data Bank to Facilitate Structural Data Mining

TL;DR

This paper introduces PDBMine, a reformulated version of the Protein Data Bank that significantly reduces storage requirements and enhances the ability to mine protein structural data for insights and predictions.

Contribution

The paper presents a novel transformation of the PDB that enables efficient data storage and powerful structural data mining capabilities.

Findings

Reduced storage space by an order of magnitude

Enabled exploration of amino acid sequence prevalence

Provided a mechanism for protein structure prediction

Abstract

Large scale initiatives such as the Human Genome Project, Structural Genomics, and individual research teams have provided large deposits of genomic and proteomic data. The transfer of data to knowledge has become one of the existing challenges, which is a consequence of capturing data in databases that are optimally designed for archiving and not mining. In this research, we have targeted the Protein Databank (PDB) and demonstrated a transformation of its content, named PDBMine, that reduces storage space by an order of magnitude, and allows for powerful mining in relation to the topic of protein structure determination. We have demonstrated the utility of PDBMine in exploring the prevalence of dimeric and trimeric amino acid sequences and provided a mechanism of predicting protein structure.