An Efficient Clustering Algorithm on Next-Generation Sequence Data

The clustering algorithms are an unsupervised machine learning methodology widely utilized in various fields to find and identify patterns in data. In bioinformatics, clustering plays a crucial role that primarily helps to find relations between similar gene sequences. In this work, a new clustering technique has been proposed that uses the concept of tensors, Riemannian manifolds, and geodesics to cluster a given set of biological data. This new method of clustering Next-Generation Sequence data has been developed keeping in mind that it should yield previously undetected patterns in such type of data. The proposed work is compared with well-known clustering techniques such as k-means, DBSCAN, hierarchical and fuzzy c-means clustering algorithms via several metrics (DB index, Dunn index, Rand score, and Silhouette score) and shows its efficiency.