FamPlex: a resource for entity recognition and relationship resolution of human protein families and complexes in biomedical text mining

Abstract

BackgroundFor automated reading of scientific publications to extract useful information about molecular mechanisms it is critical that genes, proteins and other entities be correctly associated with uniform identifiers, a process known as named entity linking or “grounding.” Correct grounding is essential for resolving relationships among mined information, curated interaction databases, and biological datasets. The accuracy of this process is largely dependent on the availability of machine-readable resources associating synonyms and abbreviations commonly found in biomedical literature with uniform identifiers.ResultsIn a task involving automated reading of ∼215,000 articles using the REACH event extraction software we found that grounding was disproportionately inaccurate for multi-protein families (e.g., “AKT”) and complexes with multiple subunits (e.g.“NF- κB”). To address this problem we constructed FamPlex, a manually curated resource defining protein families and complexes as they are commonly encountered in biomedical text. In FamPlex the gene-level constituents of families and complexes are defined in a flexible format allowing for multi-level, hierarchical membership. To create FamPlex, text strings corresponding to entities were identified empirically from literature and linked manually to uniform identifiers; these identifiers were also mapped to equivalent entries in multiple related databases. FamPlex also includes curated prefix and suffix patterns that improve named entity recognition and event extraction. Evaluation of REACH extractions on a test corpus of ∼54,000 articles showed that FamPlex significantly increased grounding accuracy for families and complexes (from 15 to 71%). The hierarchical organization of entities in FamPlex also made it possible to integrate otherwise unconnected mechanistic information across families, subfamilies, and individual proteins. Applications of FamPlex to the TRIPS/DRUM reading system and the Biocreative VI Bioentity Normalization Task dataset demonstrated the utility of FamPlex in other settings.ConclusionFamPlex is an effective resource for improving named entity recognition, grounding, and relationship resolution in automated reading of biomedical text. The content in FamPlex is available in both tabular and Open Biomedical Ontology formats at https://github.com/sorgerlab/famplex under the Creative Commons CC0 license and has been integrated into the TRIPS/DRUM and REACH reading systems.

GeneWalk identifies relevant gene functions for a biological context using network representation learning

“…The GeneWalk applications in this study used the INDRA 15,16 and Pathway Commons 14 knowledge bases which enable automated assembly of a GeneWalk network. Although these databases are optimized for human genes, we show that when mouse genes can be mapped unambiguously to their human orthologues, a network can still be assembled.…”

“…Here, represent the input weights of w I , which constitute the vector representations used for our GeneWalk analysis, and the output weights for w O . For the vector dimensionality d, we tested different values (2,3,4,6,8,12,16,32,50,500), and found that d=8 was optimal because the variance of the resulting cosine similarity distributions was largest, indicating the highest sensitivity of detection of similarity between node pairs. Lower dimensionality generally resulted in high similarity between all nodes, whereas higher dimensionality lowered all similarity values; both cases resulted in a reduced variability.…”

“…We calculated the log-likelihood ratio in Python and utilized the scipy.stats.chi2.sf function to determine the p-value of our test statistic. 15,16 or Pathway Commons 14,21 knowledge base, all molecular reactions in which these genes are involved, are retrieved and assembled in a condition-specific gene regulatory network, to which GO ontology and annotations are then connected. Through network representation learning, each gene and GO term can be represented as vectors, suitable for similarity significance testing.…”

“…For each data set described in this study, we queried the pathway databases and machine reading results from REACH and Sparser (run on all Medline abstracts and PubMedCentral manuscripts) for interactions involving the DE genes in the dataset. The resultant set of statements consisted only of relationships among DE genes, GO terms, and protein families and complexes containing the DE genes, obtained from the FamPlex ontology 16 . The final set of statements was then used as an input to the core GeneWalk algorithm as described below.…”

Assembling a corpus of phosphoproteomic annotations using ProtMapper to normalize site information from databases and text mining

“…Text mining extracts information not only about kinases that directly phosphorylate a target site, but also kinases that lie further upstream as well as non-kinase regulators (e.g., growth factors). REACH and Sparser also use FamPlex identifiers, a taxonomy of protein families and complexes for text mining, to extract and normalize phosphorylation events expressed in terms of kinase classes, e.g., "ERK", "AKT", "AMPK" [20]. For an equal comparison between text mining and databases, we therefore repeated the comparison by restricting the regulators to specific human kinase proteins and found that text mining still yielded a substantial body of new information, with 3,792 unique human kinase-site pairs reported by machine readers and not PhosphoSitePlus, and 3,118 that did not appear in any curated database ( Fig 3C).…”