OCMR: A comprehensive framework for optical chemical molecular recognition.

Artificial intelligence (AI) has achieved significant progress in the field of drug discovery. AI-based tools have been used in all aspects of drug discovery, including chemical structure recognition. We propose a chemical structure recognition framework, Optical Chemical Molecular Recognition (OCMR), to improve the data extraction capability in practical scenarios compared with the rule-based and end-to-end deep learning models. The proposed OCMR framework enhances the recognition performances via the integration of local information in the topology of molecular graphs. OCMR handles complex tasks like non-canonical drawing and atomic group abbreviation and substantially improves the current state-of-the-art results on multiple public benchmark datasets and one internally curated dataset.

Tensile Behaviors and Strain Hardening Mechanisms in a High-Mn Steel with Heterogeneous Microstructure.

Heterogeneous structures with both heterogeneous grain structure and dual phases have been designed and obtained in a high-Mn microband-induced plasticity (MBIP) steel. The heterogeneous structures show better synergy of strength and ductility as compared to the homogeneous structures. Higher contribution of hetero-deformation induced hardening to the overall strain hardening was observed and higher density of geometrically necessary dislocations were found to be induced at various domain boundaries in the heterogeneous structures, resulting in higher extra strain hardening for the observed better tensile properties as compared to the homogeneous structures. MBIP effect is found to be still effective in the coarse austenite grains of heterogeneous structures, while the typical Taylor lattice structure and the formation of microband are not observed in the ultra-fine austenite grains of heterogeneous structures, indicating that decreasing grain size might inhibit the occurrence of microbands. High density of dislocation is also observed in the interiors of BCC grains, indicating that both phases are deformable and can accommodate plastic deformation. It is interesting to note that the deformation mechanisms are highly dependent on the phase and grain size for the present MBIP steel with heterogeneous structures.

Fingerprint-based computational models of 5-lipo-oxygenase activating protein inhibitors: Activity prediction and structure clustering.

Inflammatory diseases can be treated by inhibiting 5-lipo-oxygenase activating protein (FLAP). In this study, a data set containing 2,112 FLAP inhibitors was collected. A total of 25 classification models were built by five machine learning algorithms with five different types of fingerprints. The best model, which was built by support vector machine algorithm with ECFP_4 fingerprint had an accuracy and a Matthews correlation coefficient of 0.862 and 0.722 on the test set, respectively. The predicted results were further evaluated by the application domain dSTD-PRO (a distance between one compound to models). Each compound had a dSTD-PRO value, which was calculated by the predicted probabilities obtained from all 25 models. The application domain results suggested that the reliability of predicted results depended mainly on the compounds themselves rather than algorithms or fingerprints. A group of customized 10-bit fingerprint was manually defined for clustering the molecular structures of 2,112 FLAP inhibitors into eight subsets by K-Means. According to the clustering results, most of inhibitors in two subsets (subsets 2 and 4) were highly active inhibitors. We found that aryl oxadiazole/oxazole alkanes, biaryl amino-heteroarenes, two aromatic rings (often N-containing) linked by a cyclobutene group, and 1,2,4-triazole group were typical fragments in highly active inhibitors.

Classification of Cyclooxygenase-2 Inhibitors Using Support Vector Machine and Random Forest Methods.

This work reports the classification study conducted on the biggest COX-2 inhibitor data set so far. Using 2925 diverse COX-2 inhibitors collected from 168 pieces of literature, we applied machine learning methods, support vector machine (SVM) and random forest (RF), to develop 12 classification models. The best SVM and RF models resulted in MCC values of 0.73 and 0.72, respectively. The 2925 COX-2 inhibitors were reduced to a data set of 1630 molecules by removing intermediately active inhibitors, and 12 new classification models were constructed, yielding MCC values above 0.72. The best MCC value of the external test set was predicted to be 0.68 by the RF model using ECFP_4 fingerprints. Moreover, the 2925 COX-2 inhibitors were clustered into eight subsets, and the structural features of each subset were investigated. We identified substructures important for activity including halogen, carboxyl, sulfonamide, and methanesulfonyl groups, as well as the aromatic nitrogen atoms. The models developed in this study could serve as useful tools for compound screening prior to lab tests.

SAR study on inhibitors of GIIA secreted phospholipase A2 using machine learning methods.

GIIA secreted phospholipase A2 (GIIA sPLA2 ) is a potent target for drug discovery. To distinguish the activity level of the inhibitors of GIIA sPLA2 , we built 24 classification models by three machine learning algorithms including support vector machine (SVM), decision tree (DT), and random forest (RF) based on 452 compounds. The molecules were represented by CORINA descriptors, MACCS fingerprints, and ECFP4 fingerprints, respectively. The dataset was split into a training set containing 312 compounds and a test set containing 140 compounds by Kohonen's self-organizing map (SOM) strategy and a random strategy. A recursive feature elimination (RFE) method and an information gain (IG) method were used in the selection of molecular descriptors. Three favorable performing models were obtained. They were built by SVM algorithm with CORINA descriptors (Models 1A and 2A) and ECFP4 fingerprints (Model 10A). In the prediction of test set of Model 10A, the accuracy reached 90.71%, and the Matthews correlation coefficient (MCC) values reached 0.82. In addition, the 452 inhibitors were clustered into eight subsets by K-Means algorithm for analyzing their structural features. It was found that highly active inhibitors mainly contained indole scaffold or indolizine scaffold and four side chains.