Limits...
Phenome-driven disease genetics prediction toward drug discovery.

Chen Y, Li L, Zhang GQ, Xu R - Bioinformatics (2015)

Bottom Line: Our approach achieved significantly improved performance over a baseline method, which used only one phenotype data source.We also found literature evidence to support a number of drugs among the top 200 candidates.In summary, we demonstrated that a novel strategy combining unique disease phenotype data with system approaches can lead to rapid drug discovery. nlp. edu/public/data/DMN

View Article: PubMed Central - PubMed

Affiliation: Department of Electrical Engineering and Computer Science, Department of Epidemiology and Biostatistics and Department of Family Medicine and Community Health, Case Western Reserve University, Cleveland, OH 44106, USA.

Show MeSH

Related in: MedlinePlus

Average AUCs of de novo gene prediction for our approach (green) and the baseline approach (blue). We compared overall AUCs, as well as the AUCs when the numbers of false positive genes are up to 10, 50, 100, 300, 500 and 1000
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4542779&req=5

btv245-F3: Average AUCs of de novo gene prediction for our approach (green) and the baseline approach (blue). We compared overall AUCs, as well as the AUCs when the numbers of false positive genes are up to 10, 50, 100, 300, 500 and 1000

Mentions: Our approach is effective in de novo gene predictions, and outperforms the baseline method by boosting the phenotype knowledge. Specifically, our method achieves an average AUC of 90.33%, which is significantly higher than 81.28% for the baseline method using mimMiner alone (P < e−12). Figure 3 shows that at six false positive cutoffs, integrating DMN and mimMiner achieves significantly higher AUCs (P < e−18) than using only mimMiner. For example, at the cutoff of 10, we achieve an average AUC of 59.19%, while that for the baseline method is 24.17% (P < e−95). For the diseases that only have one associated gene in OMIM, our method successfully predicted the tested genes in top 1 for 52.12% of diseases, while the baseline method succeeded in 11.47% prioritizations (P < e−4). These results show that de novo gene prediction highly depends on disease phenotype relationships, and our method successfully took the advantage of more comprehensive knowledge in multiple phenotypic networks to achieve better performance.Fig. 3.


Phenome-driven disease genetics prediction toward drug discovery.

Chen Y, Li L, Zhang GQ, Xu R - Bioinformatics (2015)

Average AUCs of de novo gene prediction for our approach (green) and the baseline approach (blue). We compared overall AUCs, as well as the AUCs when the numbers of false positive genes are up to 10, 50, 100, 300, 500 and 1000
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4542779&req=5

btv245-F3: Average AUCs of de novo gene prediction for our approach (green) and the baseline approach (blue). We compared overall AUCs, as well as the AUCs when the numbers of false positive genes are up to 10, 50, 100, 300, 500 and 1000
Mentions: Our approach is effective in de novo gene predictions, and outperforms the baseline method by boosting the phenotype knowledge. Specifically, our method achieves an average AUC of 90.33%, which is significantly higher than 81.28% for the baseline method using mimMiner alone (P < e−12). Figure 3 shows that at six false positive cutoffs, integrating DMN and mimMiner achieves significantly higher AUCs (P < e−18) than using only mimMiner. For example, at the cutoff of 10, we achieve an average AUC of 59.19%, while that for the baseline method is 24.17% (P < e−95). For the diseases that only have one associated gene in OMIM, our method successfully predicted the tested genes in top 1 for 52.12% of diseases, while the baseline method succeeded in 11.47% prioritizations (P < e−4). These results show that de novo gene prediction highly depends on disease phenotype relationships, and our method successfully took the advantage of more comprehensive knowledge in multiple phenotypic networks to achieve better performance.Fig. 3.

Bottom Line: Our approach achieved significantly improved performance over a baseline method, which used only one phenotype data source.We also found literature evidence to support a number of drugs among the top 200 candidates.In summary, we demonstrated that a novel strategy combining unique disease phenotype data with system approaches can lead to rapid drug discovery. nlp. edu/public/data/DMN

View Article: PubMed Central - PubMed

Affiliation: Department of Electrical Engineering and Computer Science, Department of Epidemiology and Biostatistics and Department of Family Medicine and Community Health, Case Western Reserve University, Cleveland, OH 44106, USA.

Show MeSH
Related in: MedlinePlus