The membranes were then stripped and re-probed with desired antibodies as a loading control

process. 2 / 17 Improving Detection of Drug-Drug Interactions in Pharmacovigilance Fig 1. Flowchart with the different steps implicated in the study. doi:10.1371/journal.pone.0129974.g001 Methods DDI reference standard We collected a reference standard with 149 DDIs present in the intersection of both DrugBank and Veterans Association Hospital database. The collected DDIs produced the effect of arrhythmias and related terms, such as QT prolongation or increased heart rhythm. In our reference standard there are DDIs with different levels of documentation, from “well established through controlled studies” to “theoretical interactions but pharmacological reasons lead clinicians to recognize the possible interaction”. The 149 DDI pairs comprised 162 drugs and were included in a 162162 drug-drug matrix MedChemExpress CSP-1103 called M1. We codified the 149 reference standard DDIs in M1 PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19703425 with value 1 in each respective cell, and the non-DDIs with value 0. 3 / 17 Improving Detection of Drug-Drug Interactions in Pharmacovigilance Fig 2. Flowchart including the steps implicated in the calculation of different similarity measures. Drugs were represented as fingerprints, i.e. bit vector codifying the presence or absence of structural keys, adverse effects, targets, drug-drug interactions or ATC codes. The Tanimoto coefficient between all the fingerprint pairs is calculated and placed in a drug-drug similarity matrix. Different M2 matrices are calculated weighted with the different similarity measures. doi:10.1371/journal.pone.0129974.g002 Drug similarity-based calculation To calculate drug similarity we used different measures. Fig 2 shows the workflow used to calculate some similarity measures. A detailed explanation about the construction of drug similarity-based models can be found in previous publications. Different drug similarity matrices were generated at this step. 2D molecular structure drug similarity. We calculated MACCS fingerprints for all the 162 drugs in our reference standard. MACCS represents the 2D molecular structure as a vector that codifies the presence or absence of different structural keys or sub-fragments. A detailed description of the PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19704080 fingerprint calculation can be found in 4 / 17 Improving Detection of Drug-Drug Interactions in Pharmacovigilance previous publications. We compared pairs of MACCS fingerprints using the Tanimoto coefficient. The Tc is the ratio between the number of features in the intersection and the union of two fingerprints. The Tc ranges from 0 to 1, which means minimum and maximum similarity respectively. Once we calculated the Tc for all the drug pairs, we constructed a 162162 drug similarity matrix. In each cell of the matrix we placed the Tc for the drug pair. 3D molecular structure drug similarity. We downloaded from DrugBank the isomeric SMILES codes of the 162 drugs in our reference standard. Isomeric SMILES codes provide information about the chemical structure but also allow the specification of the configuration of chiral centers. We pre-processed the database using the LigPrep module in the Schrdinger 2011 package. Through this process, when there are non-specified chiral centers in some drugs, a maximum of three enantiomers was generated. We performed Monte Carlo Multiple Minimum conformational analysis calculations using Macromodel to determine the most stable 3D molecular structure for each drug. We retained the structure with the minimum potential energy OPLS_2005 as a drug-template for the next shape screenin