Computational drug repurposing to predict approved and novel drug-disease associations

Khalid, Zoya; Sezerman, Uğur

doi:10.1016/j.jmgm.2018.08.005

Cited by 11 publications

(11 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is highly recommended that, when developing novel algorithms, one should not only concentrate on novelty but also identify a good feature dataset 39 . Therefore, we examined our training dataset and determined the threshold for 90% accuracy required for machine learning algorithms to succeed.…”

Section: Discussionmentioning

confidence: 99%

THREE STEPS NOVEL HARD MARGIN ENSEMBLE MACHINE LEARNING METHOD CLASSIFIES UNCERTAINMEFVGENE VARIANTS

Alay

Demir²,

Kirişçi

2023

Preprint

View full text Add to dashboard Cite

Introduction: The International Study Group for Systemic Autoinflammatory Diseases (INSAID) consensus criteria revealed that the clinical outcomes of more than half of the MEFV gene variants are uncertain. We aimed to detect more accurate classifications of MEFV variants while simultaneously reducing MEFV variant uncertainty. Material-Methods: We extracted variants of the MEFV gene from the infevers database. We then determined the optimal number of in silico instruments for our model. On the training dataset, we implemented seven machine learning algorithms on MEFV gene variants with known clinical effects. We evaluated the effectiveness of our model in three steps: First, we performed machine-learning algorithms on the training dataset and implemented those with a prediction accuracy of greater than 90 percent. Second, we compared our gene-level and protein-level prediction results. Finally, we compared our prediction results to clinical outcomes. Results: Our analysis included 266 of 381 MEFV gene variants and four computational tools (Revel, SIFT, MetaLR, and FATHMM). In our training dataset, the accuracy of three machine learning algorithms (RF: 100%, CRAT: 100%, and KNN: 91%) exceeded the threshold value. Thus, the dataset contained 134 likely pathogenic (LP) variants and 132 likely benign (LB) variants. We found that B30.2 domain variants were 2.5 times more likely to be LP than LB (χ2:12.693, p < 0.001, OR: 2.595 [1.532-4.132]. Discussion: Considering that the clinical effects of 60% of MEFV gene variants have not yet been determined, a combined evaluation of our methods and patients' clinical manifestations significantly simplifies the interpretation of unknown variants.

show abstract

Section: Discussionmentioning

confidence: 99%

THREE STEPS NOVEL HARD MARGIN ENSEMBLE MACHINE LEARNING METHOD CLASSIFIES UNCERTAINMEFVGENE VARIANTS

Alay

Demir²,

Kirişçi

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The second main approach by which these DFs could aid in the discovery of new antibacterial compounds is drug repurposing. There is extensive literature in which using QSAR models to reposition approved drugs is posed as a key tool in the development of new drug-disease associations [24,25]. More specifically, this approach has been already successfully used to identify antibacterial activity in several drugs approved for different purposes (Table 8) [26][27][28][29][30][31].…”

Section: Discussionmentioning

confidence: 99%

Molecular Topology for the Discovery of New Broad-Spectrum Antibacterial Drugs

et al. 2020

View full text Add to dashboard Cite

In this study, molecular topology was used to develop several discriminant equations capable of classifying compounds according to their antibacterial activity. Topological indices were used as structural descriptors and their relation to antibacterial activity was determined by applying linear discriminant analysis (LDA) on a group of quinolones and quinolone-like compounds. Four equations were constructed, named DF1, DF2, DF3, and DF4, all with good statistical parameters such as Fisher–Snedecor’s F (over 25 in all cases), Wilk’s lambda (below 0.36 in all cases) and percentage of correct classification (over 80% in all cases), which allows a reliable extrapolation prediction of antibacterial activity in any organic compound. From the four discriminant functions, it can be extracted that the presence of sp3 carbons, ramifications, and secondary amine groups in a molecule enhance antibacterial activity, whereas the presence of 5-member rings, sp2 carbons, and sp2 oxygens hinder it. The results obtained clearly reveal the high efficiency of combining molecular topology with LDA for the prediction of antibacterial activity.

show abstract

“…Iwata et al use AUPRC to assess internal performance of reconstructing known drug-indication associations made using supervised network inference, and compare their results to those obtained by others [70]. Khalid and Sezerman report AUPRC along with AUC and mean percentile rank to demonstrate the ability of their platform, which combines protein-protein interaction, pathway, protein binding site structural and disease similarity data, to capture known drug-indication associations [71]. With respect to cross-platform comparability, the authors applied their algorithm to evaluate performance using gold standard data available online for three other platforms and found that they obtain better AUC values using their methods but with data from other platforms [71].…”

Section: Precision and Precision-recall And Area Under The Precisionmentioning

confidence: 99%

Evaluating Performance of Drug Repurposing Technologies

Schuler

Falls

Mangione

et al. 2020

Preprint

View full text Add to dashboard Cite

Drug repurposing technologies are growing in number and maturing. However, comparison to each other and to reality is hindered due to lack of consensus with respect to performance evaluation. Such comparability is necessary to determine scientific merit and to ensure that only meaningful predictions from repurposing technologies carry through to further validation and eventual patient use. Here, we review and compare performance evaluation measures for these technologies using version 2 of our shotgun repurposing Computational Analysis of Novel Drug Opportunities (CANDO) platform to illustrate their benefits, drawbacks, and limitations. Understanding and using different performance evaluation metrics ensures robust cross platform comparability, enabling us to continuously strive towards optimal repurposing by decreasing time and cost of drug discovery and development.

show abstract

Computational drug repurposing to predict approved and novel drug-disease associations

Cited by 11 publications

References 30 publications

THREE STEPS NOVEL HARD MARGIN ENSEMBLE MACHINE LEARNING METHOD CLASSIFIES UNCERTAINMEFVGENE VARIANTS

THREE STEPS NOVEL HARD MARGIN ENSEMBLE MACHINE LEARNING METHOD CLASSIFIES UNCERTAINMEFVGENE VARIANTS

Molecular Topology for the Discovery of New Broad-Spectrum Antibacterial Drugs

Evaluating Performance of Drug Repurposing Technologies

Contact Info

Product

Resources

About