We are now showing the curated evidence used for assigning names and, if possible, gene symbols, publications, and Enzyme Commission numbers on nearly 70% (83 million) microbial RefSeq proteins. This evidence includes a hierarchical collection of curated Hidden Markov Model (HMM)-based and BLAST-based protein families, and conserved domain architectures.
The latest version of the Conserved Domain Database contains 3,272 new or updated NCBI-curated domains and now mirrors Pfam version 31 as well as models from NCBIfams, a collection of protein family hidden Markov models (HMMs) for improving bacterial genome annotation. A fine-grained classification of the major facilitator superfamily has also been added. You can find this updated content on the CDD FTP site.
If you’re a protein researcher, one thing you may want to do is to find homologs for a protein of interest on the basis of its sequence. This can provide insights into what the protein does and how it does it, and may identify proteins with known three-dimensional structures that can serve as models for the protein of interest. The Conserved Domains Database (CDD) groups proteins that have strong sequence similarity to protein domain fingerprints and allows you to search these groups with any protein sequence. Such searches are often more sensitive than standard BLAST searches since the scoring matrices used are tuned to locate important functional sites and sequence motifs that are highly conserved within the domain. You can then use the results to explore the evolutionary relationships of these proteins or identify these important sequence and structural features.
Here is a method to find protein sequences from many organisms that contain a particular conserved domain: