Makoto Taiji

A 281 Tflops calculation for X-ray protein structure analysis with special-purpose computers MDGRAPE-3

We have achieved a sustained calculation speed of 281 Tflops for the optimization of the 3-D structures of proteins from the X-ray experimental data by the Genetic Algorithm - Direct Space (GA-DS) method. In this calculation we used MDGRAPE-3, special-purpose computer for molecular simulations, with the peak performance of 752 Tflops. In the GA-DS method, a set of selected parameters which define the crystal structures of proteins is optimized by the Genetic Algorithm. As a criterion to estimate the model parameters, we used the reliability factor R1 which indicates the statistical difference between the calculated and the measured diffraction data. To evaluate this factor it is necessary to reconstruct the diffraction patterns of the model structures every time the model is updated. Therefore, in this method the nonequispaced Discrete Fourier Transformation (DFT) used to calculate the diffraction patterns dominates most of the computation time. To accelerate DFT calculations, we used the special-purpose computer, MDGRAPE-3. A molecule, Carbamoyl-Phosphate Synthetase was investigated. The final reliability factors were much smaller than the typical values obtained in other methods such as the Molecular Replacement (MR) method. Our results successfully demonstrate that high-performance computing with GA-DS method on special-purpose computers is effective for the structure determination of biological molecules and the method has a potential to be widely used in near future.

Use of the Multilayer Fragment Molecular Orbital Method to Predict the Rank Order of Protein–Ligand Binding Affinities: A Case Study Using Tankyrase 2 Inhibitors

In computational drug discovery, ranking a series of compound analogues in the order that is consistent with the experimental binding affinities remains a challenge. Many of the computational methods available for evaluating binding affinities have adopted molecular mechanics (MM)-based force fields, although they cannot completely describe protein–ligand interactions. By contrast, quantum mechanics (QM) calculations play an important role in understanding the protein–ligand interactions; however, their huge computational costs hinder their application in drug discovery. In this study, we have evaluated the ability to rank the binding affinities of tankyrase 2 ligands by combining both MM and QM calculations. Our computational approach uses the protein–ligand binding energies obtained from a cost-effective multilayer fragment molecular orbital (MFMO) method combined with the solvation energy obtained from the MM-Poisson–Boltzmann/surface area (MM-PB/SA) method to predict the binding affinity. This approach enabled us to rank tankyrase 2 inhibitor analogues, outperforming several MM-based methods, including rescoring by molecular docking and the MM-PB/SA method alone. Our results show that this computational approach using the MFMO method is a promising tool for predicting the rank order of the binding affinities of inhibitor analogues.

GRAPE project for a dedicated tera-flops computer

We are constructing a one tera-flops machine dedicated to astronomical many-body problems. It consists of parallelized GRAPE machines connected to a host workstation. The GRAPE machines only calculate forces between particles in the system by pipeline architecture. We designed and fabricated LSI chips for it, and about 2000 chips are being connected in parallel. The machine will be in operation by summer of 1995. General concept and features of the machine, mode of parallelization, and their merits are discussed in addition to scientific objectives of the project.<<ETX>>

Molecular Dynamics Machine: Highly Parallelized Special-Purpose-Computer for Molecular Dynamics Simulations

We are now developing Molecular Dynamics Machine (MDM) for classical molecular dynamics simulations. The target performance is 100 Tflops which will enable us to perform a million particles molecular dynamics simulation without truncating the Coulomb force using Ewald method. Its predicted performance in real application is about 30 Tflops and it can calculate 106 time-steps of MD simulation with a million particles in about a day. Total system will be completed in the year 2000.