Combining structural and bioactivity-based fingerprints improves prediction performance and scaffold hopping capability

Laufkötter, Oliver; Sturm, Noé; Bajorath, Jürgen; Chen, Hongming; Engkvist, Ola

doi:10.1186/s13321-019-0376-1

Cited by 37 publications

(32 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These identifiers are then usually hashed to a bit vector with a fixed length in order to allow the comparison of different representations. In this work, we set a vector length of 1024 bits using the Morgan FPs implementation of RDKit [15] since it was found that the use of a larger vector size with ML algorithms such as random forest could only allow negligible performance improvements [17].…”

Section: Molecular Fingerprintsmentioning

confidence: 99%

VenomPred: A Machine Learning Based Platform for Molecular Toxicity Predictions

Galati

Stefano

Martinelli

et al. 2022

IJMS

View full text Add to dashboard Cite

The use of in silico toxicity prediction methods plays an important role in the selection of lead compounds and in ADMET studies since in vitro and in vivo methods are often limited by ethics, time, budget and other resources. In this context, we present our new web tool VenomPred, a user-friendly platform for evaluating the potential mutagenic, hepatotoxic, carcinogenic and estrogenic effects of small molecules. VenomPred platform employs several in-house Machine Learning (ML) models developed with datasets derived from VEGA QSAR, a software that includes a comprehensive collection of different toxicity models and has been used as a reference for building and evaluating our ML models. The results showed that our models achieved equal or better performance than those obtained with the reference models included in VEGA QSAR. In order to improve the predictive performance of our platform, we adopted a consensus approach combining the results of different ML models, which was able to predict chemical toxicity better than the single models. This improved method was thus implemented in the VenomPred platform, a freely accessible webserver that takes the SMILES (Simplified Molecular-Input Line-Entry System) strings of the compounds as input and sends the prediction results providing a probability score about their potential toxicity.

show abstract

Section: Molecular Fingerprintsmentioning

confidence: 99%

VenomPred: A Machine Learning Based Platform for Molecular Toxicity Predictions

Galati

Stefano

Martinelli

et al. 2022

IJMS

View full text Add to dashboard Cite

show abstract

“… 16 In other words, the PhGs shared by diverse active compounds can be useful for scaffold hopping (SH). 8 , 9 , 17 − 19 SH requires a method to capture the functional similarity with a focus on the interaction and freedom from scaffold-based or structural similarity. 8 , 9 …”

Section: Introductionmentioning

confidence: 99%

Visualization of Topological Pharmacophore Space with Graph Edit Distance

Nakano

Miyao

2022

ACS Omega

View full text Add to dashboard Cite

A topological pharmacophore (TP) is a chemical graph-based pharmacophore representation, where nodes are pharmacophoric features (PF) and edges are topological distances between PFs. Previously proposed sparse pharmacophore graphs (SPhGs) for TPs were shown to be effective in identifying structurally different active compounds while maintaining the interpretability of the graphs. However, one limitation of using SPhGs as queries is that many structurally similar SPhGs can be identified from a set of active compounds, requiring the classification and visualization of SPhGs, followed by an understanding of the pharmacophore hypotheses. In this study, we propose a scheme for SPhG analysis based on dimensionality reduction techniques with the graph edit distance (GED) metric. This metric enables measuring similarities among SPhGs in a quantitative manner. The visualization of SPhGs, which themselves are the graphs shared by active compounds, can help us understand the pharmacophore hypotheses as well as the data set. As a proof-of-concept study, we generated two-dimensional SPhG-maps using three dimensionality reduction techniques for six biological targets. A comparison with other pharmacophore representations was also conducted. We demonstrated knowledge extraction (interpretation of the data set) from the generated maps. Our findings include a suitable mapping algorithm as well as a pharmacophore hypothesis analysis procedure using an SPhG-map.

show abstract

“…The key goal of QSAR studies is to determine a mathematical model between the property under investigation, and one or more molecular descriptors [12][13][14]. Using the model, similar bioactivities of compounds not involved in the training set can be predicted from their structural descriptors [15]. Here, we used genetic function approximation (GFA) and ELM models to predict the bioactivity of the molecules under investigation.…”

Section: Introductionmentioning

confidence: 99%

Predicting the bioactivity of 2-alkoxycarbonylallyl esters as potential antiproliferative agents against pancreatic cancer (MiaPaCa-2) cell lines: GFA-based QSAR and ELM-based models with molecular docking

Oyeneyin

Obadawo

Olanrewaju

et al. 2021

Journal of Genetic Engineering and Biotechnology

View full text Add to dashboard Cite

Background The number of cancer-related deaths is on the increase, combating this deadly disease has proved difficult owing to resistance and some serious side effects associated with drugs used to combat it. Therefore, scientists continue to probe into the mechanism of action of cancer cells and designing novel drugs that could combat this disease more safely and effectively. Here, we developed a genetic function approximation model to predict the bioactivity of some 2-alkoxyecarbonyl esters and probed into the mode of interaction of these molecules with an epidermal growth factor receptor (3POZ) using the three-dimensional quantitative structure activity relationship (QSAR), extreme learning machine (ELM), and molecular docking techniques. Results The developed QSAR model with predicted (R2pred) of 0.756 showed that the model was fit to be validated parameter for a built model and also proved that the developed model could be used in practical situation, R2 for training set (0.9929) and test set (0.8397) confirmed that the model could successfully predict the activity of new compounds due to its correlation with the experimental activity, the models generated with ELM models showed improved prediction of the activity of the molecules. The lead compounds (22 and 23) had binding energies of −6.327 and −7.232 kcalmol−1 for 22 and 23 respectively and displayed better inhibition at the binding sites of 3POZ when compared with that of the standard drug, chlorambucil (−6.0 kcalmol−1). This could be attributed to the presence of double bonds and the α-ester groups. Conclusion The QSAR and ELM models had good prognostic ability and could be used to predict the bioactivity of novel anti-proliferative drugs.

show abstract

Combining structural and bioactivity-based fingerprints improves prediction performance and scaffold hopping capability

Cited by 37 publications

References 21 publications

VenomPred: A Machine Learning Based Platform for Molecular Toxicity Predictions

VenomPred: A Machine Learning Based Platform for Molecular Toxicity Predictions

Visualization of Topological Pharmacophore Space with Graph Edit Distance

Predicting the bioactivity of 2-alkoxycarbonylallyl esters as potential antiproliferative agents against pancreatic cancer (MiaPaCa-2) cell lines: GFA-based QSAR and ELM-based models with molecular docking

Contact Info

Product

Resources

About