LIDeB tools


🖐About Us

We are a drug discovery team with an interest in the development of publicly available open-source customizable cheminformatics tools to be used in computer-assisted drug discovery. We belong to the Laboratory of Bioactive Research and Development (LIDeB) of the National University of La Plata (UNLP), Argentina. Our research group is focused on computer-guided drug repurposing and rational discovery of new drug candidates to treat epilepsy and neglected tropical diseases.

 

⚡ Coming soon more webapps! 🎈


 


📌Heatmaps- Similarity

ℹ About this App

Build a Heatmap of molecular similarity. These plots of inter-molecular similarity (computed as Tanimoto similarity coefficient using Morgan fingerprints and other molecular fingerprinting systems) allow for a fast, visual inspection of the molecular diversity of the datasets, and also preliminary detection of clusters within a dataset. The resulting plots are downloadable as .png files through a simple right click on your mouse!.



   

 

 

📌LuDe

ℹ About this App

LUDe (LIDEB’s Useful Decoys) is WebApp that generates, from a set of active compounds, decoys (putative inactive compounds) which can be used to retrospectively validate virtual screening tools/protocols. Decoys are molecules that have not been tested against a molecular target of interest but due to their structural features are presumably not prone to bind the target with high affinity. LUDe finds decoys in a curated ChEMBL27 database which are paired with the known active compounds in relation to certain general physicochemical properties (e.g., molecular weight, log P, and others) but which are topologically different from the query compounds. LUDe is conceptually similar to the Directory of Useful Decoys enhanced, but additional filters have been serially implemented to assure the topological dissimilarity between the decoys and the active compounds.

   

📌iRaPCA-Clustering

ℹ About this App

iRaPCA Clustering is based on an iterative combination of the random subspace approach (feature bagging), dimensionality reduction through Principal Component Analysis (PCA) and the k-means algorithm. The optimal number of clusters k and the best subset of descriptors are selected from plots of silhouette coefficient against different k values and subsets. Different validation metrics can be downloaded once the process is finished. A number of graphs may be built and readily downloaded through a simple click.



   

 

📌SOMoC Clustering

ℹ About this App

SOMoC is a clustering methodology based on the combination of molecular fingerprinting, dimensionality reduction by the Uniform Manifold Approximation and Projection (UMAP) algorithm and clustering with the Gaussian Mixture Model (GMM) algorithm.

 

 

 

📌Metrics

ℹ About this App

Metrics to evaluate the performance of the classificatory models Web App to evaluate the perfomance of classificatory models by calculation of AUCROC, BEDROC, Accuracy, F-measure, PR and EF. Additionaly, using the generated PPV surfaces you can select an adequate score threshold for prospective virtual screenings. The tool uses the following packages RDKIT, Scikit-learn, Plotly

   

📌LISTo

ℹ About this App

LIDeB’s Standardization Tool is a WebApp to standardize SMILES based in MolVS. The tool uses the following packages: RDKIT, MolVS, mols2grid Default setting will perform next actions to each smiles: #Remove explicit hydrogens by RemoveHs (RDKIT) #Kekulize, check valencies, set aromaticity, conjugation and hybridization by SanitizeMol (RDKIT) #Disconnect metal atoms that are defined as covalently bonded to non-metals by MetalDisconnector (RDKIT) #Correct functional groups and recombining charges by Normalizer (RDKIT) #Fix charges and reionize a molecule by Reionizer (RDKIT) #Select the largest covalent fragment by fragment_parent (MolVS) #Remove stereochemistry by stereo_parent (MolVS) #Select the more stable tautomeric form by tautomer_parent (MolVS) #Remove charges – neutralization – by charge_parent (MolVS) #Select the most abundant isotope by isotope_parent (MolVS)