Boruta is a feature selection algorithm that works with random forests. For each feature, it generates a shadow feature by permutation and compares the importance of the original feature to the maximum importance of the shadow features. Features are deemed important if their importance exceeds the maximum shadow importance. It repeats this process iteratively until all features are either deemed important or unimportant. The document provides an example application of Boruta for detecting important k-mers in DNA sequences that are indicative of aptamers, and finds it performs well at selecting meaningful features.