Supervised learning with word embeddings derived from PubMed captures latent knowledge about protein kinases and cancer

Ravanmehr, Vida; Blau, Hannah; Cappelletti, Luca; Fontana, Tommaso; Carmody, Leigh C.; Coleman, Benjamin; George, Joshy; Reese, Justin; Joachimiak, Marcin P.; Bocci, Giovanni; Hansen, Peter J.; Bult, Carol J.; Rueter, Jens; Casiraghi, Elena; Valentini, Giorgio; Mungall, Christopher J.; Oprea, Tudor I.; Robinson, Peter N.

doi:10.1093/nargab/lqab113

Cited by 5 publications

(3 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This first implementation returns data from Ravenmehr et al. ( 49 ), which predicts a relationship between specific cancers and kinases. Incorporating data to Pharos allows scientists to have access to all the visualizations and subset analysis tools that Pharos offers, for their own datasets.…”

Section: Discussionmentioning

confidence: 99%

Pharos 2023: an integrated resource for the understudied human proteome

Kelleher

Sheils

Mathias

et al. 2022

Nucleic Acids Research

Self Cite

View full text Add to dashboard Cite

The Illuminating the Druggable Genome (IDG) project aims to improve our understanding of understudied proteins and our ability to study them in the context of disease biology by perturbing them with small molecules, biologics, or other therapeutic modalities. Two main products from the IDG effort are the Target Central Resource Database (TCRD) (http://juniper.health.unm.edu/tcrd/), which curates and aggregates information, and Pharos (https://pharos.nih.gov/), a web interface for fusers to extract and visualize data from TCRD. Since the 2021 release, TCRD/Pharos has focused on developing visualization and analysis tools that help reveal higher-level patterns in the underlying data. The current iterations of TCRD and Pharos enable users to perform enrichment calculations based on subsets of targets, diseases, or ligands and to create interactive heat maps and UpSet charts of many types of annotations. Using several examples, we show how to address disease biology and drug discovery questions through enrichment calculations and UpSet charts.

show abstract

Section: Discussionmentioning

confidence: 99%

Pharos 2023: an integrated resource for the understudied human proteome

Kelleher

Sheils

Mathias

et al. 2022

Nucleic Acids Research

Self Cite

View full text Add to dashboard Cite

show abstract

“…The corpus used to test our proposal is composed of 10,584,195 abstracts and titles published between January, 2010 and November, 2020 and available in PubMed; they were downloaded from the FTP site of the National Center for Biotechnology Information (NCBI) by using Marea, a software 13 1 that automatically parses the annual baseline and daily file-updates, p rovided i n t he f orm o f m etadata by N CBI, a nd extract the PubMed ID and year of publication of each paper.…”

Section: Methodsmentioning

confidence: 99%

Replacing non-biomedical concepts improves embedding of biomedical concepts

Niyonkuru,

Soto Gomez,

Casiraghi

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

ObjectivesConcept embeddings are low-dimensional vector representations of concepts such as MeSH:D009203 (Myocardial Infarction), whose similarity in the embedded vector space reflects their semantic similarity. Here, we test the hypothesis that non-biomedical concept synonym replacement can improve the quality of biomedical concepts embeddings.Materials and methodsWe developed an approach that leverages WordNet to replace sets of synonyms with the most common representative of the synonym set.ResultsWe tested our approach on 1055 concept sets and found that, on average, the mean intracluster distance was reduced by 8% in the vector-space. Assuming that homophily of related concepts in the vector space is desirable, our approach tends to improve the quality of embeddings.Discussion and ConclusionThis pilot study shows that non-biomedical synonym replacement tends to improve the quality of embeddings of biomedical concepts using the Word2Vec algorithm. We have implemented our approach in a freely available Python package available athttps://github.com/TheJacksonLaboratory/wn2vec.

show abstract

“…Despite these efforts, nearly 164 out of the 534 known kinases remain relatively understudied and are referred to as “dark” kinases by the NIH Illuminating the Druggable Genome consortium (IDG) ( Berginski et al, 2021 ; Gillespie et al, 2022 ; Kelleher et al, 2023 ). Characterizing the functions of these dark kinases is crucial because they work in conjunction with other well-studied kinases in signaling pathways and are also frequently mutated, or abnormally expressed, in human diseases such as cancers ( Berginski et al, 2021 ; Brognard & Hunter, 2011 ; Collins et al, 2018 ; Ravanmehr et al, 2021 ; Soleymani et al, 2022 ). While considerable progress has been made in illuminating the functions of several dark kinases, placing these kinases in a pathway or a cell signaling network context remains a major bioinformatics challenge.…”

Section: Introductionmentioning

confidence: 99%

Predicting protein and pathway associations for understudied dark kinases using pattern-constrained knowledge graph embedding

Salcedo,

Gravel,

Keshavarzi

et al. 2023

PeerJ

View full text Add to dashboard Cite

The 534 protein kinases encoded in the human genome constitute a large druggable class of proteins that include both well-studied and understudied “dark” members. Accurate prediction of dark kinase functions is a major bioinformatics challenge. Here, we employ a graph mining approach that uses the evolutionary and functional context encoded in knowledge graphs (KGs) to predict protein and pathway associations for understudied kinases. We propose a new scalable graph embedding approach, RegPattern2Vec, which employs regular pattern constrained random walks to sample diverse aspects of node context within a KG flexibly. RegPattern2Vec learns functional representations of kinases, interacting partners, post-translational modifications, pathways, cellular localization, and chemical interactions from a kinase-centric KG that integrates and conceptualizes data from curated heterogeneous data resources. By contextualizing information relevant to prediction, RegPattern2Vec improves accuracy and efficiency in comparison to other random walk-based graph embedding approaches. We show that the predictions produced by our model overlap with pathway enrichment data produced using experimentally validated Protein-Protein Interaction (PPI) data from both publicly available databases and experimental datasets not used in training. Our model also has the advantage of using the collected random walks as biological context to interpret the predicted protein-pathway associations. We provide high-confidence pathway predictions for 34 dark kinases and present three case studies in which analysis of meta-paths associated with the prediction enables biological interpretation. Overall, RegPattern2Vec efficiently samples multiple node types for link prediction on biological knowledge graphs and the predicted associations between understudied kinases, pseudokinases, and known pathways serve as a conceptual starting point for hypothesis generation and testing.

show abstract

Supervised learning with word embeddings derived from PubMed captures latent knowledge about protein kinases and cancer

Cited by 5 publications

References 49 publications

Pharos 2023: an integrated resource for the understudied human proteome

Pharos 2023: an integrated resource for the understudied human proteome

Replacing non-biomedical concepts improves embedding of biomedical concepts

Predicting protein and pathway associations for understudied dark kinases using pattern-constrained knowledge graph embedding

Contact Info

Product

Resources

About