Papers vs. PDB: Well-studied proteins that lack structural information

Papers vs. PDB shows proteins that are mentioned in at least 10 papers yet are not similar to any protein of known structure.

Filtering:

Only the 100 most-cited proteins were shown.

Technical Details

• A protein is considered to lack structural information if at least half of the protein has no homolog in PDB with ≥30% average amino acid identity. Furthermore this region must be at least 50 amino acids long.
• Homologs were identified using protein BLAST against a database of PDB sequences clustered at 90%, using the options -e 0.01 -m "F S". Regions with no hits were confirmed by blasting against the entire PDB database with the option -e 0.01.
• As of May 2018, 27,408 proteins are linked to at least 10 papers. 7,541 of these lack structural information and 3,851 lack structural information over their entire length.
• Transmembrane helices (TMHs) were identified with TMHMM.
• A protein is considered partly soluble if it has one or more TMHs but a region of at least 50 amino acids has no TMHs.
• Average disorder was predicted using IUPredL.
• A protein is considered disordered if the average probability of disorder is above 40%.
• Hits to PFam were computed using PFam-A. HMMer 3.1b2, and the gathering cutoff (--cut_ga).
• The organism selector includes all organisms with 50 or more entries. These organisms cover 90% of all entries.
• This analysis is from January 10, 2020.

Downloads

• All uncovered well-cited proteins (tab-delimited)

Developed by Morgan Price, Arkin lab, LBL