ed with docking DOCK6 ZDOCK/PIPER/Hex MolDock DARC PLANTS AutoDock Vina GPUperTrAmber GPU enabled functionality Molecular dynamics Amber scoring Fast Fourier Transforms Initially only scoring, then also differential evolution Simultaneously scoring multiple particles Concurrent grid-based search Runs docking concurrently from different starting orientations Scoring very large systems by decomposition Speedup 236 6.56 156 276 276 606 626 1006 Docking methods have been adapted for GPU computing using a variety of strategies. These require different degrees of CPU-GPU communication, and accordingly enable varying LY3039478 speedups relative to the analogous CPU-only protocol. doi:10.1371/journal.pone.0070661.t001 8 Fast Docking on GPUs via Ray-Casting in turn suggests that the current framework could be adapted by increasing the size of the problem allocated to the GPU at each iteration. Through further careful examination of the relationship between the number of particles and the number of iterations, it may prove possible to achieve equivalent convergence more quickly more particles and fewer iterations. Alternatively, further parallelization may be realized by bundling particles corresponding to different ligand conformers for simultaneous scoring on the GPU, rather than carry out separate optimization of each conformer. The fact that additional calculations can be likely carried out on the GPU with little additional cost also offers the opportunity to fundamentally change the DARC scoring paradigm: either by simultaneously using multiple sets of rays originating from distinct origins within the protein, and/or by adding new components to capture effects of electrostatics. In short, any enhancement that increases the computational burden per iteration that is carried by the GPU is likely to yield further speedup relative to the CPU alone. Given fixed computational resources allocated for completion of a project, the ability to carry out docking more rapidly will have profound implications for applications of DARC. In the most obvious case, this speedup will allow screening against very large libraries that previously may not have been tractable, for example the complete ZINC database or a library of hypothetical compounds likely amenable to straightforward synthesis. 17804190 Even in cases in which a relatively small library of interest is to be screened, this speedup will allow an increase in the number of conformers screened per compound; this in turn is expected to reduce the number of false negatives in the screen, by increasing the likelihood of including an active conformer. This speedup may further allow the use of multiple pre-built receptor conformations for docking, providing a means to implicitly represent receptor flexibility and thus allow further diversity in collection of hits identified. Paracetamol is an effective analgesic, a single oral dose of 1000 mg having an NNT of 3.6 for at least 50% pain reduction over 46 hours in patients with postoperative pain. However, the formation of toxic metabolites, such as N-acetyl-p-benzoquinone imine, in the liver is becoming an 7906496 increasing concern particularly during longterm use, potentially preventing the full exploitation of paracetamol’s analgesic activity. We have previously demonstrated that paracetamol is metabolized to the TRPV1 activator AM404 and that activation of TRPV1 in the brain by this metabolite contributes to the antinociceptive activity of oral paracetamol in rodents. Mapping the