ABCMdb: The mutation browser

The ABC mutations database contains mentions of ABC protein mutations extracted from the literature using the MutMiner automated data mining method.

The MutMiner pipeline was designed to recognize any mention of amino acid substitution occuring in text, including natural and experimental mutations found in genotypic or protein in vitro studies.

Mutations were extracted from the full-text version of PubMed papers resulting from keyword searches for various ABC proteins.


  1. Mutations are identified automatically and stored in the database
    After mutation extraction, the hits for ABCC6, G1, G2, G5 and G8 were verified manually to assure data quality, constituting a benchmark set. There is also an on-line manual curation function, so that registered users can mark individual mentions of mutations as correct or incorrect. The database will be updated with recent PubMed search hits regularly.
    The overall accuracy of the MutMiner system is over 95% according to our benchmark set.
    The current apparent accuracy of the system based on verified hits is 99.22%.
    The percent of hits that have not been verified yet is 95.18%.
  2. Alignments for comparative analysis of mutations
    The web application make it easy to look up mutations in and around homologous positions in various other ABC proteins (Browse – select protein – list all mutations – select mutation – homologous mutations). Alignments are provided for certain reginos of ABC proteins, but custom alignments can also be uploaded (MyProfile – alignments).
  3. Mapping the mutations in 3D
    The web applications also allows the mapping of amino acid positions onto 3D models of proteins or domains with the map mutations function (Browse – select protein – map mutations). Homologous positions are calculated using alignments. Custom PDB files with protein structures can also be uploaded (MyProfile – PDB files).

If you find our database useful, please cite: