In the quest to address Mendelian disorders, understanding and rectifying the fundamental molecular imbalance holds greater promise than merely addressing the symptoms. A recent computational tool, "parsing modifiers via article annotations" (PARMESAN), has been developed to expedite the identification of potential treatments. By sifting through PubMed and PubMed Central, PARMESAN assembles a vast knowledge base containing critical drug-gene and gene-gene relationships.
The rapid rise in next-generation sequencing technologies has spotlighted thousands of genes linked to Mendelian disorders. However, even after pinpointing these genes, many related diseases are managed only symptomatically. The absence of therapies targeting the molecular imbalance is a significant clinical gap. An efficient way to bridge this gap is by having a comprehensive understanding of the disease's molecular pathways and identifying drugs that modulate these pathways appropriately.
Several databases, including BioGRID, STRING, and BioPlex, have been essential in for inferring gene-gene relationships by mapping protein-protein interactions. Meanwhile, databases like Reactome, KEGG, NeuroGeM, and PhenoModifier, curated meticulously through expert literature reviews, offer insights into the directionality of these relationships. While the same level of care is applied in maintaining drug-gene databases such as DrugBank and DGIdb, the sheer volume of relevant publications makes timely manual updates challenging. Over 2 million such publications are added yearly.
Tools like PubTator, SemRep, and BioBERT have tried to automate the extraction of valuable information from this vast pool of medical literature. However, tools like BioBERT require substantial time, and others, like SemMedDB, have accessibility limitations.
Enter PARMESAN. Designed to stay current with the latest in literature, this tool trawls through PubMed's 27 million abstracts and PubMed Central's 5 million full-text articles. PARMESAN aims to decipher the relationships between genes and between drugs and genes. This research demonstrated PARMESAN's precision and its potential in earmarking novel therapeutic candidates for genetic disorders.
The process of validating a drug's therapeutic efficacy for a specific disorder can be both resource-intensive and prolonged. For disorders where protein imbalance is the culprit –either due to insufficient protein or toxic overproduction—PARMESAN can be invaluable in guiding drug discovery. Its predictive capabilities improve as the literature expands, capturing new gene-gene and drug-gene relationships. Although a centralized database for new discoveries remains ideal, PARMESAN's automated literature curation serves as a strong alternative.
The tool's reliability was underscored by the consistency between its drug predictions and the DGIdb. Notably, for more than half of the disease genes reported by the UDN, PARMESAN offered drug predictions with over 91% matching directionality.
In addition to drug predictions, PARMESAN can also steer genetic modifier screens, enhancing our understanding of gene-gene relationships and aiding in therapeutic development. Although promising, PARMESAN isn't without limitations. Its dependency on available literature could be restrictive, especially for new disease genes. Also, verifying the absence of relationships between two entities remains a challenge, affecting PARMESAN's predictive accuracy for uncharted relationships. Lastly, while the tool currently focuses on upstream regulators, predicting downstream treatments could further amplify its therapeutic potential.
Despite these constraints, PARMESAN's capabilities in accelerating the literature review process and pinpointing test-worthy candidate therapeutics for previously untreatable genetic disorders cannot be understated.
Read the original publication in the American Journal of Human Genetics.
The rapid rise in next-generation sequencing technologies has spotlighted thousands of genes linked to Mendelian disorders. However, even after pinpointing these genes, many related diseases are managed only symptomatically. The absence of therapies targeting the molecular imbalance is a significant clinical gap. An efficient way to bridge this gap is by having a comprehensive understanding of the disease's molecular pathways and identifying drugs that modulate these pathways appropriately.
Several databases, including BioGRID, STRING, and BioPlex, have been essential in for inferring gene-gene relationships by mapping protein-protein interactions. Meanwhile, databases like Reactome, KEGG, NeuroGeM, and PhenoModifier, curated meticulously through expert literature reviews, offer insights into the directionality of these relationships. While the same level of care is applied in maintaining drug-gene databases such as DrugBank and DGIdb, the sheer volume of relevant publications makes timely manual updates challenging. Over 2 million such publications are added yearly.
Tools like PubTator, SemRep, and BioBERT have tried to automate the extraction of valuable information from this vast pool of medical literature. However, tools like BioBERT require substantial time, and others, like SemMedDB, have accessibility limitations.
Enter PARMESAN. Designed to stay current with the latest in literature, this tool trawls through PubMed's 27 million abstracts and PubMed Central's 5 million full-text articles. PARMESAN aims to decipher the relationships between genes and between drugs and genes. This research demonstrated PARMESAN's precision and its potential in earmarking novel therapeutic candidates for genetic disorders.
The process of validating a drug's therapeutic efficacy for a specific disorder can be both resource-intensive and prolonged. For disorders where protein imbalance is the culprit –either due to insufficient protein or toxic overproduction—PARMESAN can be invaluable in guiding drug discovery. Its predictive capabilities improve as the literature expands, capturing new gene-gene and drug-gene relationships. Although a centralized database for new discoveries remains ideal, PARMESAN's automated literature curation serves as a strong alternative.
The tool's reliability was underscored by the consistency between its drug predictions and the DGIdb. Notably, for more than half of the disease genes reported by the UDN, PARMESAN offered drug predictions with over 91% matching directionality.
In addition to drug predictions, PARMESAN can also steer genetic modifier screens, enhancing our understanding of gene-gene relationships and aiding in therapeutic development. Although promising, PARMESAN isn't without limitations. Its dependency on available literature could be restrictive, especially for new disease genes. Also, verifying the absence of relationships between two entities remains a challenge, affecting PARMESAN's predictive accuracy for uncharted relationships. Lastly, while the tool currently focuses on upstream regulators, predicting downstream treatments could further amplify its therapeutic potential.
Despite these constraints, PARMESAN's capabilities in accelerating the literature review process and pinpointing test-worthy candidate therapeutics for previously untreatable genetic disorders cannot be understated.
Read the original publication in the American Journal of Human Genetics.