Identifying Gene-Specific Variations in Biomedical Text

Abstract

The influence of genetic variations on diseases or cellular processes is the main focus of many investigations, and results of biomedical studies are often only accessible through scientific publications. Automatic extraction of this information requires recognition of the gene names and the accompanying allelic variant information. In a previous work, the OSIRIS system for the detection of allelic variation in text based on a query expansion approach was communicated. Challenges associated with this system are the relatively low recall for variation mentions and gene name recognition. To tackle this challenge, we integrate the ProMiner system developed for the recognition and normalization of gene and protein names with a conditional random field (CRF)-based recognition of variation terms in biomedical text. Following the newly developed normalization of variation entities, we can link textual entities to Single Nucleotide Polymorphism database (dbSNP) entries. The performance of this novel approach is evaluated, and improved results in comparison to state-of-the-art systems are reported.

Challenges in the association of human single nucleotide polymorphism mentions with unique database identifiers

“…Another application useful for the conversion of different SNP description is SNP-converter [42]. More and more publications describe SNPs also in terms of dbSNP accession numbers [43], which is supported by the latest mutation nomenclature. For example, the mention rs2306220:A>G is a valid SNP description.…”

“…Approaches relying on machine learning methods commonly identify wildtype, mutated allele and location separately [43,57]. These approaches require subsequent association with extracted sub-entities (alleles and location) to build a complete SNP tuple.…”

“…In such cases, the SNP can be unambiguously identified and mapped to dbSNP. In a previous experiment [43], such mentions have been extracted with the regular expression “[rR][sS][ ]*[1–9][0–9]*”. On a test set of 300 extracted mentions, a recall of 100 % and a comparatively small precision of 74 % has been achieved.…”

“…Some consider sentence boundaries [29], while other approaches associate a SNP with all extracted genes and try to validate the results using the dbSNP database [43,53]. Another approach introduced the “graph bigram association algorithm” for the purpose of building the required protein-variation pairs [58].…”

Minimum tag error for discriminative training of conditional random fields

“…Because CRFs are basically defined as conditional models of label sequences given observation sequences, they can make use of flexible overlapping features and overcome label bias problems. In recent years, CRFs have been successfully applied to many tasks, such as gene identification [11], spoken language understanding (SLU) [10], part-of-speech (POS) tagging [14], name entity recognition (NER) [3,17], and shallow parsing [23], especially Chinese word segmentation [6,21].…”