Benchmarking Active Learning Protocols for Ligand-Binding Affinity Prediction.

Active learning (AL) has become a powerful tool in computational drug discovery, enabling the identification of top binders from vast molecular libraries. To design a robust AL protocol, it is important to understand the influence of AL parameters, as well as the features of the data sets on the outcomes. We use four affinity data sets for different targets (TYK2, USP7, D2R, Mpro) to systematically evaluate the performance of machine learning models [Gaussian process (GP) model and Chemprop model], sample selection protocols, and the batch size based on metrics describing the overall predictive power of the model (R2, Spearman rank, root-mean-square error) as well as the accurate identification of top 2%/5% binders (Recall, F1 score). Both models have a comparable Recall of top binders on large data sets, but the GP model surpasses the Chemprop model when training data are sparse. A larger initial batch size, especially on diverse data sets, increased the Recall of both models as well as overall correlation metrics. However, for subsequent cycles, smaller batch sizes of 20 or 30 compounds proved to be desirable. Furthermore, adding artificial Gaussian noise to the data up to a certain threshold still allowed the model to identify clusters with top-scoring compounds. However, excessive noise (<1σ) did impact the model's predictive and exploitative capabilities.

From Proteins to Ligands: Decoding Deep Learning Methods for Binding Affinity Prediction.

Accurate in silico prediction of protein-ligand binding affinity is important in the early stages of drug discovery. Deep learning-based methods exist but have yet to overtake more conventional methods such as giga-docking largely due to their lack of generalizability. To improve generalizability, we need to understand what these models learn from input protein and ligand data. We systematically investigated a sequence-based deep learning framework to assess the impact of protein and ligand encodings on predicting binding affinities for commonly used kinase data sets. The role of proteins is studied using convolutional neural network-based encodings obtained from sequences and graph neural network-based encodings enriched with structural information from contact maps. Ligand-based encodings are generated from graph-neural networks. We test different ligand perturbations by randomizing node and edge properties. For proteins, we make use of 3 different protein contact generation methods (AlphaFold2, Pconsc4, and ESM-1b) and compare these with a random control. Our investigation shows that protein encodings do not substantially impact the binding predictions, with no statistically significant difference in binding affinity for KIBA in the investigated metrics (concordance index, Pearson's R Spearman's Rank, and RMSE). Significant differences are seen for ligand encodings with random ligands and random ligand node properties, suggesting a much bigger reliance on ligand data for the learning tasks. Using different ways to combine protein and ligand encodings did not show a significant change in performance.

SkiNet: A deep learning framework for skin lesion diagnosis with uncertainty estimation and explainability.

Skin cancer is considered to be the most common human malignancy. Around 5 million new cases of skin cancer are recorded in the United States annually. Early identification and evaluation of skin lesions are of great clinical significance, but the disproportionate dermatologist-patient ratio poses a significant problem in most developing nations. Therefore a novel deep architecture, named as SkiNet, is proposed to provide faster screening solution and assistance to newly trained physicians in the process of clinical diagnosis of skin cancer. The main motive behind SkiNet's design and development is to provide a white box solution, addressing a critical problem of trust and interpretability which is crucial for the wider adoption of Computer-aided diagnosis systems by medical practitioners. The proposed SkiNet is a two-stage pipeline wherein the lesion segmentation is followed by the lesion classification. Monte Carlo dropout and test time augmentation techniques have been employed in the proposed method to estimate epistemic and aleatoric uncertainty. A novel segmentation model named Bayesian MultiResUNet is used to estimate the uncertainty on the predicted segmentation map. Saliency-based methods like XRAI, Grad-CAM and Guided Backprop are explored to provide post-hoc explanations of the deep learning models. The ISIC-2018 dataset is used to perform the experimentation and ablation studies. The results establish the robustness of the proposed model on the traditional benchmarks while addressing the black-box nature of such models to alleviate the skepticism of medical practitioners by incorporating transparency and confidence to the model's prediction.

DMENet: Diabetic Macular Edema diagnosis using Hierarchical Ensemble of CNNs.

Diabetic Macular Edema (DME) is an advanced stage of Diabetic Retinopathy (DR) and can lead to permanent vision loss. Currently, it affects 26.7 million people globally and on account of such a huge number of DME cases and the limited number of ophthalmologists, it is desirable to automate the diagnosis process. Computer-assisted, deep learning based diagnosis could help in early detection, following which precision medication can help to mitigate the vision loss. METHOD: In order to automate the screening of DME, we propose a novel DMENet Algorithm which is built on the pillars of Convolutional Neural Networks (CNNs). DMENet analyses the preprocessed color fundus images and passes it through a two-stage pipeline. The first stage detects the presence or absence of DME whereas the second stage takes only the positive cases and grades the images based on severity. In both the stages, we use a novel Hierarchical Ensemble of CNNs (HE-CNN). This paper uses two of the popular publicly available datasets IDRiD and MESSIDOR for classification. Preprocessing on the images is performed using morphological opening and gaussian kernel. The dataset is augmented to solve the class imbalance problem for better performance of the proposed model. RESULTS: The proposed methodology achieved an average Accuracy of 96.12%, Sensitivity of 96.32%, Specificity of 95.84%, and F-1 score of 0.9609 on MESSIDOR and IDRiD datasets. CONCLUSION: These excellent results establish the validity of the proposed methodology for use in DME screening and solidifies the applicability of the HE-CNN classification technique in the domain of biomedical imaging.