ERpred: a web server for the prediction of subtype-specific estrogen receptor antagonists

Abstract

Estrogen receptors alpha and beta (ERα and ERβ) are responsible for breast cancer metastasis through their involvement of clinical outcomes. Estradiol and hormone replacement therapy targets both ERs, but this often leads to an increased risk of breast and endometrial cancers as well as thromboembolism. A major challenge is posed for the development of compounds possessing ER subtype specificity. Herein, we present a large-scale classification structure-activity relationship (CSAR) study of inhibitors from the ChEMBL database which consisted of an initial set of 11,618 compounds for ERα and 7,810 compounds for ERβ. The IC50 was selected as the bioactivity unit for further investigation and after the data curation process, this led to a final data set of 1,593 and 1,281 compounds for ERα and ERβ, respectively. We employed the random forest (RF) algorithm for model building and of the 12 fingerprint types, models built using the PubChem fingerprint was the most robust (Ac of 94.65% and 92.25% and Matthews correlation coefficient (MCC) of 89% and 76% for ERα and ERβ, respectively) and therefore selected for feature interpretation. Results indicated the importance of features pertaining to aromatic rings, nitrogen-containing functional groups and aliphatic hydrocarbons. Finally, the model was deployed as the publicly available web server called ERpred at http://codes.bio/erpred where users can submit SMILES notation as the input query for prediction of the bioactivity against ERα and ERβ.

StackER: a novel SMILES-based stacked approach for the accelerated and efficient discovery of ERα and ERβ antagonists

“…As mentioned above, ERpred 21 is the only SMILE notation-based approach for predicting ERα and ERβ inhibitors. Since ERpred was not developed based on the up-to-date dataset constructed herein, we implemented ERpred using the same procedure as mentioned in the previous study.…”

“…Then, we obtained the subsequent dataset consisting of 2532 and 1577 compounds for ERα and ERβ, respectively. To generate active and inactive compounds, we applied the same criteria as employed in previous studies 21 , 24 – 26 . As a result, we obtained actives and inactives (ERα and ERβ) of (1145 and 736) and (851 and 471), respectively.…”

“…Then, all the features were ranked according to their importance scores. The RF method is widely applied in various biological and chemical classification problems 21 , 24 , 39 , 40 . After obtaining the ranked features, we constructed n feature sets containing the m top-ranked important features ranging from top m start to m end with an interval of s. The values of m start , m end , s , and n depend on the feature dimension.…”

“…It's important to note that most of these research endeavors primarily focus on targeting ERα rather than ERβ 20 . To date, there is only one ML-based approach (named ERpred 21 ) that is developed for predicting the effectiveness of inhibitors against ERα and ERβ. ERpred is a random forest-based model trained on 659 and 714 compounds for ERα and ERβ, respectively.…”

Machine-learning-based on multimodality radiomics analysis for the Preoperative Prediction for local relapse in osteosarcoma

“…RF also provides a measure of feature importance, allowing for the identi cation of signi cant radiomics features that contribute to the prediction and enhancing model interpretability. [25] These advantages make RF particularly suitable for small-sample radiomics studies with limited samples and complex relationships between features and outcomes. The RF had best predictive e ciency (AUCs: training set 0.999 vs. test set 0.868) of four machine learning models in our study.…”