This simple rule correctly identified 75.0% of the 28 speedy human targets and only unexpectedly misclassified 13.2% of 53 non-speedy human targets. misclassification of some human targets by this simple rule. Investigation and knowledge of trial-speed differentiating features enable prioritized drug discovery and development. database [38], developers public releases and literatures. The first-in-class drugs were selected based on the descriptions in the FDA annual Novel Drug Summary (2011C17) and the literatures [39], wherein the drugs were explicitly described as the first-in-class or Diphenyleneiodonium chloride of the new therapeutic mechanism. For each drug with single reported target in the searched source, that target was tentatively recorded as its efficacy target. For each drug without target information or with 2 targets, its efficacy target(s) was (were) searched by literatures on the basis that it (they) is usually (are) linked to the clinically tested therapeutics using the established criterion of efficacy targets (Supplementary Table S1). For multi-target drugs, the primary efficacy targets were selected based on their essential roles in the targeted disease(s) or the availability of other drugs in the advanced development stages against the same target and the same disease(s). Drugs of the innovative targets in clinical trials during 1981C2017 All clinical trial drugs during 1981C2017 were collected from such resources (Supplementary Table S2) as the PhRMA medicines in development reports, the drug pipeline reports, annual reports and announcements of 371 companies and 18 research organizations, therapeutic target database (TTD) [7], IUPHAR/BPS Guide to Pharmacology [40] and the reputable commercial databases (MDL? Drug Data Report 2004, CenterWatch Drugs in Clinical Trials 2007, Thomson Reuters Pharma? 2010 and 2012, Springer AdisInsight 2015). Their annual trial status Rabbit polyclonal to IQCD in the period of 1981C2017 was further searched or verified against the records in the NIH ClinicalTrials.gov [38], developers public releases and literatures. Synonyms reported in these sources were recorded and the duplicates were further removed. For each drug with 2 clinical trials within the same year, the highest trial phase was recorded for that Diphenyleneiodonium chloride year. These drugs were compared with the approved drugs in the official website of Drugs@FDA, popular databases (TTD, Drugbank and ChEMBL) and literatures to remove those approved before 2004. The targets of these drugs were searched and compared to the innovative targets of the first-in-class drugs approved during 2004C17 to select drug targeting one or more of these innovative targets. These drugs were subsequently used to Diphenyleneiodonium chloride determine the trial phase or approval status of each innovative target in every year in 1981C2017. For targets with 2 drugs, the highest phase or approval status was recorded for that year. Druggability characteristics of the innovative targets The disease role of each target was searched from the literatures by keyword combinations between target names and `disease, `therapeutics, `mechanism or `function. The drug-binding site features were searched from the literatures by keyword combinations between target names and `binding site or `binding. The relationship between the drug-binding domain of each target to that of pre-existing targets was determined as follows: (1) sequence similarity between the drug-binding domain of the target and the pre-existing targets was computed using BLAST [41] with E-value of 0.001 and 0.001C0.05 for high and fair similarity, respectively; (2) the affiliation of each target to drug-binding domain name family of pre-existing drugs (drug approved before year 2004) was assigned by condition that the target is in a Pfam family [42] that contains 1 pre-existing target (without an approved drugbefore 2004). The human systems features for the on-target collateral effects were determined as following: (1) the human protein-network of 8698 proteins and 45 932 protein-pairs was constructed based on the data of the STRING Database [43] with a confidence score of 0.95 [43C46], which was used for computing human biological network descriptors for each target based on Cytoscape [47]. For the protein-pairs with one protein paired to multiple subtypes of another protein, only one.