Sparse Proteomics Analysis – a compressed sensing-based approach for feature selection and classification of high-dimensional proteomics mass spectrometry data

Abstract

BackgroundHigh-throughput proteomics techniques, such as mass spectrometry (MS)-based approaches, produce very high-dimensional data-sets. In a clinical setting one is often interested in how mass spectra differ between patients of different classes, for example spectra from healthy patients vs. spectra from patients having a particular disease. Machine learning algorithms are needed to (a) identify these discriminating features and (b) classify unknown spectra based on this feature set. Since the acquired data is usually noisy, the algorithms should be robust against noise and outliers, while the identified feature set should be as small as possible.ResultsWe present a new algorithm, Sparse Proteomics Analysis (SPA), based on the theory of compressed sensing that allows us to identify a minimal discriminating set of features from mass spectrometry data-sets. We show (1) how our method performs on artificial and real-world data-sets, (2) that its performance is competitive with standard (and widely used) algorithms for analyzing proteomics data, and (3) that it is robust against random and systematic noise. We further demonstrate the applicability of our algorithm to two previously published clinical data-sets.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-017-1565-4) contains supplementary material, which is available to authorized users.

Comparison of marker selection methods for high throughput scRNA-seq data

“…In this work we introduce a feature selection algorithm that takes steps to mitigate this issue. This method, which we call R ANK C ORR , is inspired by the method in [13] and relies on the ranking the scRNA-seq counts before marker selection. Below, we provide some intuition as to why ranking scRNA-seq data is a useful strategy for understanding scRNA-seq counts data.…”

“…Below, we provide some intuition as to why ranking scRNA-seq data is a useful strategy for understanding scRNA-seq counts data. The method in [13] is based on convex optimization (which would generally be quite slow on large data sets); it is possible to find a quick deterministic solution to this optimization, however, so that R ANK C ORR runs quickly. R ANK C ORR is able to handle over one million cells in an amount of time that is competitive with simple statistical methods.…”

“…The rows of an scRNA-seq data matrix X do not look like standard Gaussian vectors and, importantly, do not have mean 0. In order to apply the results of [17] to sparse biological data, therefore, the authors of [13] propose standardizing the columns of X before running the optimization (4). That is, they create X std by and then let be the solution generated from X std and τ .…”

Prostate cancer recognition based on mass spectrometry sensing data and data fingerprint recovery

“…Data reduction algorithms will be of critical importance in medicine going forward, having extensive application in the areas of disease risk assessment and personalized medicine. Major efforts are focused on improving classification while reducing computation [1]. Many algorithms have been proposed to classify MS data.…”

“…al [5] propose a Recursive Feature Elimination (SVMRFE) algorithm that selects important genes/biomarkers for the classification of noisy data. The sparse proteomics analysis (SPA) is another way to complete feature selection based on the compressive sensing concept [6]. Sparse features are a small subset of features that can be used to accurately predict unknown proteomic data.…”