Density Peak Clustering of UHGP50 (DPCfam-UHGP50)

Density Peak Clustering of UHGP50 (DPCfam-UHGP50) clusters protein sequences from the Unified Human Gastrointestinal Proteome (UHGP50) into Metaclusters using the DPCfam algorithm to enable structural and functional annotation of metagenomic protein regions.

Key Features:

• Unsupervised clustering: The DPCfam algorithm groups homologous protein regions without prior labels into density-based Metaclusters.
• Metacluster formation: Each Metacluster comprises seed sequence regions that represent putative members of a protein family.
• Profile-HMM construction: Seed regions from Metaclusters are used to build profile-HMMs for sensitive identification of homologous regions across proteins.
• Multi-domain architecture grouping: Clustering captures multi-domain architectures by assembling homologous regions from multi-domain proteins into Metaclusters.
• Enhanced family coverage: The DPCfam-UHGP50 dataset contains 10,778 putative protein families, increasing coverage at both the protein and residue levels relative to Pfam.

Scientific Applications:

• Metagenomics research: Provides comprehensive classification of UHGP50 sequences to support studies of human gut microbial communities.
• Structural and functional annotation: Supplies seed regions and profile-HMMs to aid annotation of protein structure and function in metagenomic datasets.
• Medical and biological insights: Facilitates identification of protein families relevant to disease mechanisms and therapeutic research.

Methodology:

Sequence clustering using the DPCfam algorithm groups homologous regions into Metaclusters, and seed sequence regions from each Metacluster are used to construct profile-HMMs.