Figure 4. Illustration of the process for extracting residue mentions and mapping them to physical residues in protein structures.

Text mining for residue mentions is performed on the abstracts from the primary references cited in the PDB. A text residue is represented by a residue number and one or more 3-letter amino acid codes. An amino acid for a text residue is marked with an asterisk if a text occurrence suggests it is a mutation from wild type. Physical residues corresponding to a text residue are indicated using a PDB ID, chain identifier, residue number, and 3-letter amino acid code.