netDx: Software for building interpretable patient classifiers by multi-'omic data integration using patient similarity networks

Abstract

Patient classification based on clinical and genomic data will further the goal of precision medicine. Interpretability is of particular relevance for models based on genomic data, where sample sizes are relatively small (in the hundreds), increasing overfitting risk netDx is a machine learning method to integrate multi-modal patient data and build a patient classifier. Patient data are converted into networks of patient similarity, which is intuitive to clinicians who also use patient similarity for medical diagnosis. Features passing selection are integrated, and new patients are assigned to the class with the greatest profile similarity. netDx has excellent performance, outperforming most machine-learning methods in binary cancer survival prediction. It handles missing data – a common problem in real-world data – without requiring imputation. netDx also has excellent interpretability, with native support to group genes into pathways for mechanistic insight into predictive features. The netDx Bioconductor package provides multiple workflows for users to build custom patient classifiers. It provides turnkey functions for one-step predictor generation from multi-modal data, including feature selection over multiple train/test data splits. Workflows offer versatility with custom feature design, choice of similarity metric; speed is improved by parallel execution. Built-in functions and examples allow users to compute model performance metrics such as AUROC, AUPR, and accuracy. netDx uses RCy3 to visualize top-scoring pathways and the final integrated patient network in Cytoscape. Advanced users can build more complex predictor designs with functional building blocks used in the default design. Finally, the netDx Bioconductor package provides a novel workflow for pathway-based patient classification from sparse genetic data.

Simpati: patient classifier identifies signature pathways based on similarity networks for the disease prediction

“…The classes assigned to the patients were Early or Late based on their cancer stage. We increased the challenge including the performances of the current published generic-purpose pathway-based classifiers: netDx [28] and PASNet [19]. netDx creates a database of PSNs associated to pathways for each class, applies a network fusion algorithm to produce a consensus PSN and applies GeneMANIA (state-of-art gene function prediction algorithm) for the prediction of the testing patients.…”

“…In the end, the predicted classes of the testing patients collected from all the iterations are compared to their real ones for determining the classification performances. Simpati is designed to value its ability to predict based on two measures following netDx design [28]. The first one called AUC-ROC is the area under the curve where the x-axis is the false positive rate (FPR) and the y-axis is true positive rate (TPR).…”

“…Pai et al proposes netDx [27,28] as patient classifier based on the PSN paradigm. Any available datum (e.g., age, gender, gene expression, …) is converted into a PSN.…”

“…Similarity networks can represent any datum, naturally handle missing and heterogenous data, have a history of successes in gene and protein function prediction [21][22][23] and can preserve the patient privacy by being shared in place of the sensitive raw information (topic which is growing in concern) [24][25][26]. Pai and Giudice et al propose netDx [27,28] as patient classifier based on the PSN paradigm. Any available datum (e.g., age, gender, gene expression, …) is converted into a PSN.…”

StellarPath: Hierarchical-vertical multi-omics classifier synergizes stable markers and interpretable similarity networks for patient profiling

“…The recent review by Gliozzo et al [ 9 ] lists classifiers based on the PSN paradigm. It highlights that, to date, the only published method for patient classification is netDx [ 10 , 11 ]. netDx does not perform feature selection, employs the Pearson correlation as similarity measure for patient profiles described with biological omics, and does not look for interpretable networks.…”