RampDB:

The goal of RampDB is to serve as a central repository for exploring currently known and predicting RAMP interactions. RampDB itself is a MySQL database with the following data model.


Data Model:

The data model was designed with a normalized approach to reduce redudancy, and to allow for rapid acquisition of interaction data for each result.

The database is loaded automatically with scripts which search for proteins matching a certain family, and then filters out the false positive entries through a heuristic approach. This loading is also designed to recognize updates to current entries and to update them in RampDB.

Prediction Algorithm:

The main feature of RampDB is the prediction function. This utility allows the user to input either a FASTA protein or ligand query, and predicts whether or not that query could have RAMP or RAMP-like interactions.


Proteins:

For protein queries, the first step is to do a local blast against all proteins in RampDB. The goal of this is to quickly determine the potential family of the query.

If a family is identified, the next step is to use HMMscan to compare it against a pre-built HMM profile of the RAMP interacting domain in that protein to determine the similarity. If no match is found, the query is scanned against all the prebuilt HMM profiles for each family to determine the liklihood of any interaction.


Ligands:

For ligand queries, the following search terms are acceptable:

  • Ligand Name (case insensitive)
  • Partial Ligand Name (case insensitive)
  • InChiKey

When a ligand query is submitted, the first step is to search for an exact match in RampDB. If no match is found, the query is searched against PubChem's database for 2D similarity, and those results are checked for ligands known to have RAMP interactions.

The Tanimoto Score between two ligands is used to determine their similarity. This works by dividing the intersect of the descriptors between the query and subject by the union of the descriptors.

Descriptors used: Atom count, ring count, atom sequence, bond sequence, augmented atoms, degree of connectivity, element composition, type of ring fusion

Tanimoto Score:

  • Score = C / ((QS + FS) – C)
  • C = number of descriptors shared by query and subject
  • QS = number of descriptors in the query
  • FS = number of descriptors in the subject

Citation Information:

Topaz,N., Mojib,N., Chande,A.T. et al. RampDB: a web application and database for the exploration and prediction of receptor activity modifying protein interactions. Database (2017) Vol. 2017: article ID bax067; doi:10.1093/database/bax067