FeatSeekR

FeatSeekR performs unsupervised feature selection using replicated measurements by iteratively selecting features with the highest reproducibility across replicates.

Key Features:

• Iterative feature selection: Features are selected one at a time based on highest replicate reproducibility.
• Projection-based uniqueness: After selecting a feature, the dimensions spanned by previously selected features are projected out from the data to ensure subsequent features are unique.
• Reproducibility optimization: The selected feature set maximizes both replicate reproducibility and degree of uniqueness.

Scientific Applications:

• Unsupervised feature selection: Identifying reproducible and non-redundant features from replicated measurements without requiring class labels or prior knowledge.

Methodology:

FeatSeekR iteratively selects features with the highest reproducibility across replicates, after projecting out those dimensions from the data that are spanned by the previously selected features.