8,937 research outputs found
Protocol to explain graph neural network predictions using an edge-centric Shapley value-based approach
No full textHere we present EdgeSHAPer, a workflow for explaining graph neural networks by approximating Shapley values using Monte Carlo sampling. In this protocol, we describe steps to execute Python scripts for a chemical dataset from the original publication; however, this approach is also applicable to any user-provided dataset. We also detail steps encompassing neural network training, an explanation phase, and analysis via feature mapping. For complete details on the use and execution of this protocol, please refer to Mastropietro et al. (2022).(1
Protocol to explain support vector machine predictions via exact Shapley value computation
No full textShapley values from cooperative game theory are adapted for explaining machine learning predictions. For large feature sets used in machine learning, Shapley values are approximated. We present a protocol for two techniques for explaining support vector machine predictions with exact Shapley value computation. We detail the application of these algorithms and provide ready-to-use Python scripts and custom code. The final output of the protocol includes quantitative feature analysis and mapping of important features for visualization
Compounds Activity Dopamine D2 Receptor
No full textThose data were used as dataset in the work "EdgeSHAPer: Bond-Centric Shapley Value-Based Explanation Method for Graph Neural Networks". They contain compounds with activity against the dopamine D2 receptor, filtered as stated in the paper. Here you can find a .csv containing both active (label 0) and randomly selected compounds (label 1), along with the data split into training, validation and test sets, as used in the work
Toward explainable biomedical deep learning
No full textDeep learning has been extensively utilized in the domains of bioinformatics and chemoinformatics, yielding compelling results. However, neural networks have predominantly been regarded as black boxes, characterized by internal mechanisms that hinder interpretability due to the highly nonlinear functions they learn. In the biomedical field, this lack of interpretability is undesirable, as it is imperative for scientists to comprehend the reasons behind the occurrence of specific diseases or the molecular properties that make a compound effective against a particular target protein. Consequently, the inherent closure of those models keeps their results far from being trusted. To address this issue and make deep learning suitable for bioinformatics and chemoinformatics tasks, there is the urge to develop techniques for explainable artificial intelligence (XAI). These techniques should be capable of measuring the significance of input features for predictions or determining the strength of their interactions. The ability to provide explanations must be integrated into the biomedical deep learning pipeline, which utilizes available data sources to uncover new insights regarding potentially disease-associated genes, thereby facilitating the repurposing and development of new drugs. In line with this objective, this thesis focuses on the development of innovative explainability techniques for neural networks and demonstrates their effective applications in bioinformatics and medicinal chemistry. The devised models find their place in the pipeline, wherein each component of the protocol generates effective and explainable results. These results span from the discovery of disease genes to the repurposing and development of drugs. However, deep learning lives in synergy with classical machine learning models and network-based algorithms, which remain relevant in this field and, therefore, hold a place within this thesis. Moreover, they offer the basis for proper training of deep learning models and pave the way for the development of XAI techniques for neural networks. The proposed work demonstrates how XAI can benefit biomedicine, proving deep learning to be a powerful tool to solve biomedical problems and that the obtained results can be explained. This contributes to the delivery of not only accurate but also trustworthy results, fulfilling the need for explainability of medical doctors, geneticists, and scientists in the life sciences and leading toward a fully explainable biomedical deep learning pipeline
Compounds with Activity against the Dopamine D2 Receptor
No full textThe compound data sets originating from ChEMBL and its partitions were in the study "EdgeSHAPer: Bond-Centric Shapley Value-Based Explanation Method for Graph Neural Networks". They contain compounds with activity against the dopamine D2 receptor and randomly selected active compounds, respectively, curated as described in the paper. The .csv file contains both active (label 0) and randomly selected compounds (label 1), together with partitions into training, validation, and test sets used in the study
Learning characteristics of graph neural networks predicting protein–ligand affinities
No full textIn drug design, compound potency prediction is a popular machine learning application. Graph neural networks (GNNs) predict ligand affinity from graph representations of protein–ligand interactions typically extracted from X-ray structures. Despite some promising findings leading to claims that GNNs can learn details of protein–ligand interactions, such predictions are also controversially viewed. For example, evidence has been presented that GNNs might not learn protein–ligand interactions but memorize ligand and protein training data instead. We have carried out affinity predictions with six GNN architectures on community-standard datasets and rationalized the predictions using explainable artificial intelligence. The results confirm a strong influence of ligand—but not protein—memorization during GNN learning and also show that some GNN architectures increasingly prioritize interaction information for predicting high affinities. Thus, while GNNs do not comprehensively account for protein–ligand interactions and physical reality, depending on the model, they balance ligand memorization with learning of interaction patterns
Calculation of exact Shapley values for explaining support vector machine models using the radial basis function kernel
No full textMachine learning (ML) algorithms are extensively used in pharmaceutical research. Most ML models have black-box character, thus preventing the interpretation of predictions. However, rationalizing model decisions is of critical importance if predictions should aid in experimental design. Accordingly, in interdisciplinary research, there is growing interest in explaining ML models. Methods devised for this purpose are a part of the explainable artificial intelligence (XAI) spectrum of approaches. In XAI, the Shapley value concept originating from cooperative game theory has become popular for identifying features determining predictions. The Shapley value concept has been adapted as a model-agnostic approach for explaining predictions. Since the computational time required for Shapley value calculations scales exponentially with the number of features used, local approximations such as Shapley additive explanations (SHAP) are usually required in ML. The support vector machine (SVM) algorithm is one of the most popular ML methods in pharmaceutical research and beyond. SVM models are often explained using SHAP. However, there is only limited correlation between SHAP and exact Shapley values, as previously demonstrated for SVM calculations using the Tanimoto kernel, which limits SVM model explanation. Since the Tanimoto kernel is a special kernel function mostly applied for assessing chemical similarity, we have developed the Shapley value-expressed radial basis function (SVERAD), a computationally efficient approach for the calculation of exact Shapley values for SVM models based upon radial basis function kernels that are widely applied in different areas. SVERAD is shown to produce meaningful explanations of SVM predictions
NIAPU: network-informed adaptive positive-unlabeled learning for disease gene identification
No full textGene-disease associations are fundamental for understanding disease etiology
and developing effective interventions and treatments. Identifying genes not
yet associated with a disease due to a lack of studies is a challenging task in
which prioritization based on prior knowledge is an important element. The
computational search for new candidate disease genes may be eased by
positive-unlabeled learning, the machine learning setting in which only a
subset of instances are labeled as positive while the rest of the data set is
unlabeled. In this work, we propose a set of effective network-based features
to be used in a novel Markov diffusion-based multi-class labeling strategy for
putative disease gene discovery. The performances of the new labeling algorithm
and the effectiveness of the proposed features have been tested on ten
different disease data sets using three machine learning algorithms. The new
features have been compared against classical topological and
functional/ontological features and a set of network- and biological-derived
features already used in gene discovery tasks. The predictive power of the
integrated methodology in searching for new disease genes has been found to be
competitive against state-of-the-art algorithms.Comment: This article has been accepted for publication in Bioinformatics,
Published by Oxford University Pres
- …
