Protein Database

Question

[Mery]

Hi, I need to make protein database, but I have 0 expericence in it. Can you recommend me a course? Or maybe also a paper? I have found something myself, but also wated to aske some with a bigger experience. I want to learn how to built a database, cluster proteins and set thresholds. Best regards

[dthorbur]

You'll have to give more details about the database to get a detailed answer. What is the database going to be used for? How is it different from using the existing large curated proteins databases out there?

Regarding clustering of proteins, I would look into using MMseqs2, it's probably the current gold standard for clustering sequences. If you want to cluster proteins by 3D structure, then DaliLite.v5 is a good place to start.

[Mensur Dlakic]

Generally agree with this comment. However, DaliLite is fairly slow for modern structure databases that include millions of predicted structures. There are faster options with similar sensitivity.

• https://search.foldseek.com/search
• https://github.com/steineggerlab/foldseek

[dthorbur]

You're right, FoldSeek is a much better tool. I had forgotten they included a cluster function - I've only ever used it to search databases so it skipped my mind.

Answer 1

[Elisabeth Gasteiger]

Please don't hesitate to contact the UniProt helpdesk (https://www.uniprot.org/contact) if you need help with any query that would allow you to build a suitable dataset from UniProt data.