Protein point mutations are an essential component of the evolutionary and experimental analysis of protein structure and function. While many manually curated databases attempt to index point mutations, most experimentally generated point mutations and the biological impacts of the changes are described in the peer-reviewed published literature. We describe an application, Mutation GraB (Graph Bigram), that identifies, extracts, and verifies point mutations from biomedical literature. The principal problem of point mutation extraction is to link the point mutation with its associated protein and organism of origin. Our algorithm uses a graph-based bigram traversal to identify these relevant associations and exploits the Swiss-Prot protein database to verify this information. The graph bigram method is different from other models for point mutation extraction in that it incorporates frequency and positional data of all terms in an article to drive the point mutation-protein association. Our method was tested on 589 articles describing point mutations from the G protein-coupled receptor (GPCR), tyrosine kinase, and ion channel protein families. We evaluated our graph bigram metric against a word-proximity metric for term association on datasets of full-text literature in these three different protein families. Our testing shows that the graph bigram metric achieves a higher F-measure for the GPCRs (0.79 versus 0.76), protein tyrosine kinases (0.72 versus 0.69), and ion channel transporters (0.76 versus 0.74). Importantly, in situations where more than one protein can be assigned to a point mutation and disambiguation is required, the graph bigram metric achieves a precision of 0.84 compared with the word distance metric precision of 0.73. We believe the graph bigram search metric to be a significant improvement over previous search metrics for point mutation extraction and to be applicable to text-mining application requiring the association of words.

MOTIVATION: The amount of genomic and proteomic data that is published daily in the scientific literature is outstripping the ability of experimental scientists to stay current. Reviews, the traditional medium for collating published observations, are also unable to keep pace. For some specific classes of information (e.g. sequences and protein structures), obligatory data deposition policies have helped. However, a great deal of other valuable information is spread throughout the literature hindering coherent access. We are involved in the Molecular Class-Specific Information System (MCSIS) project, a collaborative effort to design and automate the maintenance of protein family databases. The first two databases, the GPCRDB and NucleaRDB, are focused on G protein-coupled receptors (GPCRs) and nuclear hormone receptors (NRs), respectively. The main aim of the MCSIS project is to gather heterogeneous data from across a variety of electronic and literature sources in order to draw new inferences about the target protein families. RESULTS: We present a computational method that identifies and extracts mutation data from the scientific literature. We focused on the extraction of single point mutations for the GPCR and NR superfamilies. After validation by plausibility filters, the mutation data is integrated into the corresponding MCSIS where it is combined with structural and sequence information already stored in these databases. We extracted and validated 2736 true point mutations from 914 articles on GPCRs and 785 true point mutations from 1094 articles on NRs. The current version of our automated extraction algorithm identifies 49.3% of the GPCR point mutations with a specificity of 87.9%, and 64.5% of the NR point mutations with a specificity of 85.8%. MuteXt routinely analyzes 100 electronic articles in approximately 1 h.