Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Dec 4;8:889.
doi: 10.3389/fphar.2017.00889. eCollection 2017.

vNN Web Server for ADMET Predictions

Affiliations
Free PMC article

vNN Web Server for ADMET Predictions

Patric Schyman et al. Front Pharmacol. .
Free PMC article

Abstract

In drug development, early assessments of pharmacokinetic and toxic properties are important stepping stones to avoid costly and unnecessary failures. Considerable progress has recently been made in the development of computer-based (in silico) models to estimate such properties. Nonetheless, such models can be further improved in terms of their ability to make predictions more rapidly, easily, and with greater reliability. To address this issue, we have used our vNN method to develop 15 absorption, distribution, metabolism, excretion, and toxicity (ADMET) prediction models. These models quickly assess some of the most important properties of potential drug candidates, including their cytotoxicity, mutagenicity, cardiotoxicity, drug-drug interactions, microsomal stability, and likelihood of causing drug-induced liver injury. Here we summarize the ability of each of these models to predict such properties and discuss their overall performance. All of these ADMET models are publically available on our website (https://vnnadmet.bhsai.org/), which also offers the capability of using the vNN method to customize and build new models.

Keywords: ADME; QSAR; applicability domain; machine learning; online web platform; open access; toxicology.

Figures

Figure 1
Figure 1
The vNN-ADMET main page. From this page, users can run ADMET models or build their own models.
Figure 2
Figure 2
Submit ADMET predictions. On the Run ADMET Models page (top) users can upload a list of query compounds in SMILES format (lower left) or manually enter compounds by using the draw structure feature (lower right).
Figure 3
Figure 3
The ADMET predictions result page. The 15 ADMET predictions for each query molecule are presented on a separate row. Predictions based on models using a restricted applicability domain are shown in solid colors and those based on models using an unrestricted applicability domain are shown in striped colors. Users can download the results from the website into a single file.
Figure 4
Figure 4
Build a classification model. On the Build Classification Model page (top), users can upload their training data and/or draw structures. On the Build Classification Model Results page (bottom), users can interactively select/deselect different smoothing factors for comparison. The graph shows accuracy of performance on the 10-fold cross validation test at different Tanimoto distances, where smoothing factors 0.2 and 1.0 are highlighted in green and blue, respectively (strikethrough smoothing factors indicate deselected values). The coverage is shown in gray. The red circle indicates the “best” model performance based on accuracy and coverage, where the black arrows show the corresponding Tanimoto-distance threshold (d0 = 0.7) and smoothing factor (h = 0.2). Although the accuracy is reduced to 88 from 90% at d0 = 0.6, the number of compounds predicted increases from 60 to 75%, which may be worth the loss in accuracy.
Figure 5
Figure 5
Run a customized model. The first step to run a customized model is to upload the training dataset, as well as the selected Tanimoto distance and smoothing factor from Figure 4. The second step is to upload query compounds. The results can be downloaded from the Run Custom Model Results page (bottom).

Similar articles

See all similar articles

Cited by 4 articles

References

    1. Abbott N. J., Rönnbäck L., Hansson E. (2006). Astrocyte-endothelial interactions at the blood-brain barrier. Nat. Rev. Neurosci. 7, 41–53. 10.1038/nrn1824 - DOI - PubMed
    1. Adler J., Parmryd I. (2010). Quantifying colocalization by correlation: the Pearson correlation coefficient is superior to the Mander's overlap coefficient. Cytometry A 77, 733–742. 10.1002/cyto.a.20896 - DOI - PubMed
    1. Ambudkar S. V., Kimchi-Sarfaty C., Sauna Z. E., Gottesman M. M. (2003). P-glycoprotein: from genomics to mechanism. Oncogene 22, 7468–7485. 10.1038/sj.onc.1206948 - DOI - PubMed
    1. Ames B. N., Durston W. E., Yamasaki E., Lee F. D. (1973). Carcinogens are mutagens: a simple test system combining liver homogenates for activation and bacteria for detection. Proc. Natl. Acad. Sci. U.S.A. 70, 2281–2285. 10.1073/pnas.70.8.2281 - DOI - PMC - PubMed
    1. Assis D. N., Navarro V. J. (2009). Human drug hepatotoxicity: a contemporary clinical perspective. Expert Opin. Drug Metab. Toxicol. 5, 463–473. 10.1517/17425250902927386 - DOI - PubMed

LinkOut - more resources

Feedback