Discovering a sparse set of pairwise discriminating features in high dimensional data
This addresses the challenge of feature discovery in high-dimensional scientific data, such as genomics, for clustering and classification tasks, representing an incremental advance with specific domain applications.
The authors tackled the problem of extracting meaningful features from high-dimensional data by proposing an unsupervised method to identify a sparse set of features that enable linear separability of clusters in a low-dimensional subspace, applying it to single-cell RNA-seq data to identify 27 key transcription factors (out of 409) with 18 known to define cell states, leading to clear signatures of known cell types that were previously elusive.
Extracting an understanding of the underlying system from high dimensional data is a growing problem in science. Discovering informative and meaningful features is crucial for clustering, classification, and low dimensional data embedding. Here we propose to construct features based on their ability to discriminate between clusters of the data points. We define a class of problems in which linear separability of clusters is hidden in a low dimensional space. We propose an unsupervised method to identify the subset of features that define a low dimensional subspace in which clustering can be conducted. This is achieved by averaging over discriminators trained on an ensemble of proposed cluster configurations. We then apply our method to single cell RNA-seq data from mouse gastrulation, and identify 27 key transcription factors (out of 409 total), 18 of which are known to define cell states through their expression levels. In this inferred subspace, we find clear signatures of known cell types that eluded classification prior to discovery of the correct low dimensional subspace.