We have developed a classifier from evolutionary contact maps called IsItABarrel. We use IsItABarrel to identify transmembrane beta barrel sequences from a diverse set of bacterial proteomes, thereby providing a new way to answer the question "Is it a barrel?". Here we offer to provide a searchable curated database of approximately two million bacterial transmembrane beta barrels (TMBBs).
Citation
Please cite Montezano D, Bernstein R, Copeland MM, Slusky JSG. General features of transmembrane beta barrels from a large database. Proc Natl Acad Sci USA. 2023;120(29):e2220762120. doi:10.1073/pnas.2220762120
License
The IsItABarrel database and filtered search results are licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0 License).
Method
Our method is based on heuristic features extracted from contact maps created from co-evolutionary information. To predict if a protein sequence is a TMBB, we use features from the protein contact map. The map features used for our prediction are the presence of the barrel closing contacts and a minimal number of hairpins (and strands) of specific lengths.
Searching the Database
To search the database of putative TMBBs, a number of scoring criteria are available, including signal peptide prediction score obtained with the SignalP algorithm and type of signal sequence, sequence length and organism name. The sequences matching the user's search criteria will be emailed to the user as a link to a downloadable, gzip compressed, tar archive file that includes a FASTA file with all the amino acid sequences and a TAB-separated text file containing each of the records for downstream analysis.
Source code
The command line program used to generate the IsItABarrelDB is available on GitHub.