Discovery of sparse, reliable omic biomarkers with Stabl

Abstract

Adoption of high-content omic technologies in clinical studies, coupled with computational methods, has yielded an abundance of candidate biomarkers. However, translating such findings into bona fide clinical biomarkers remains challenging. To facilitate this process, we introduce Stabl, a general machine learning method that identifies a sparse, reliable set of biomarkers by integrating noise injection and a data-driven signal-to-noise threshold into multivariable predictive modeling. Evaluation of Stabl on synthetic datasets and five independent clinical studies demonstrates improved biomarker sparsity and reliability compared to commonly used sparsity-promoting regularization methods while maintaining predictive performance; it distills datasets containing 1,400–35,000 features down to 4–34 candidate biomarkers. Stabl extends to multi-omic integration tasks, enabling biological interpretation of complex predictive models, as it hones in on a shortlist of proteomic, metabolomic and cytometric events predicting labor onset, microbial biomarkers of pre-term birth and a pre-operative immune signature of post-surgical infections. Stabl is available at https://github.com/gregbellan/Stabl.

Decoding the Microbiome-Metabolome Nexus: A Systematic Benchmark of Integrative Strategies

“…However, associations found may result in artifact signals since most feature selection methods benchmarked in this paper suffer from lack of sparsity and reliability. This result is aligned with previous reports where authors have shown poor performances of traditional feature selection models [42]. Indeed, most penalized methods are mainly built upon cross-validation where small perturbations in data may yield drastic changements in results.…”

“…Indeed, most penalized methods are mainly built upon cross-validation where small perturbations in data may yield drastic changements in results. Similarly to [42] extending sparse multivariate or univariate methods to knockoff framework [43] or stability selection [44] should represent interesting avenues for improving both sparsity and reliability for compositional data [45] (Table 2).…”

“…Moreover, inspired from [42] when univariate and multivariate feature selection methods were evaluated, we considered sparsity and reliability as primary performance metrics. For univariate methods sparsity corresponds to the total number of relevant associations found by the method (here with coefficients different from zero), while reliability is the capability of a method to accurately discriminate true from false associations between two features.…”

An immune signature of postoperative cognitive decline in elderly patients

“…To identify preoperative single-cell or plasma proteomic features predictive of POCD, we employed Stabl( 41 ), a sparse machine learning method that combines multivariable predictive modeling with a data-driven feature selection process. This method is optimized for analysis of multi-omic datasets, as each omic data layer is first examined individually prior to integration into a unique predictive model.…”

Advances and Challenges in Sepsis Management: Modern Tools and Future Directions

“…Starting from 1400–35,000 features in a dataset, it extrapolates 4–34 putative biomarkers. Hence, Stabl has been applied to integrates multi-omics data of pre-term birth and a pre-operative immune signature of post-surgical infections and to predict labor onset [ 328 ]. Thus, it might be exploited in sepsis settings to predict sepsis outcomes, as well as biomarkers associated with dysfunctional immune systems.…”