Docking with Autodock and Molecular Dynamic analysis with Gromacs: Part of Indonesian Herbal Farmacological activities screening in Silico On a Cluster computing environment. Screening aktifitas farmakologis beberapa bahan aktif tumbuhan obat Indonesia secara in silico menggunakan High Performance Computing berbasis Cluster system Joint research : Arry Yanuar, Dept of Pharmacy, and Heru Suhartanto, Faculty of Computer Science, Universitas Indonesia Supported by The Indonesian Ministry of Research and Technology Office, 2009-2010,research grant Gromacs GROMACS is a versatile package program to perform molecular dynamics. GROMACS can be run with single processor or using multiple processor (parallel using standard MPI communication) Our Research is study the performance (time) between, on the Cluster computing resources and on the GPU (Graphic Processor Unit) Hastinapura Hastinapura.grid.ui.ac.id is the first Cluster computing resources, the Faculty of Computer Science Universitas Indonesia. This cluster can be used to run parallel and serial applications (gromacs). It consists of 16 dual-core machines that act as worker nodes. Hardware Specification Head node Sun Fire X2100 AMD Opteron 2.2GHz (Dual Core) 2 GB RAM Debian GNU/Linux 3.1 “Sarge” Worker nodes (16) Sun Fire X2100 AMD Opteron 2.2GHz (Dual Core) 1 GB RAM Debian GNU/Linux 3.1 “Sarge” Storage node Dual Intel Xeon 2.8GHz (HT) 2 GB RAM Debian GNU/Linux 4.0-testing “Etch” Harddisk 3x320 GB Memory Specs: Feature Support: GPU Hardware Specification Dual Core 3.2 GHz 4 GB RAM Ubuntu 9.04 64 Bit Harddisk 80 Gb Gromacs 4.05 + OpenMM GeForce GTS 250 CUDA Cores Graphics Clock (MHz) Processor Clock (MHz) Texture Fill Rate (billion/sec) 128 738 MHz 1836 MHz Memory Clock (MHz) Standard Memory Config Memory Interface Width Memory Bandwidth (GB/sec) 1100 512MB or 1 GB GDDR3 256-bit 47.2 70.4 File Preparation File Cyp34a Convert File Into .topology & .gro pdb2gmx -f 1TQN.pdb -p 1TQN.top -o 1TQN.gro Periodic Boundary Condition editconf -f 1TQN.gro -o 1TQN.gro -d 1.0 Adding solvent into the molecule genbox -cp 1TQN.gro -cs spc216.gro -p 1TQN.top –o 1TQN-solvate.pdb Energy Minimization grompp -np 16 -f md.mdp -c 1TQN.gro -p 1TQN.top -o 1TQN-md.tpr 1TQN-md.tpr is ready to be executed with 16 processor Md-job.sh #!/bin/sh # CYP34A #$ -N gromacs #$ -cwd # Jumlah prosesor #$ -pe mpich 16 #$ -l arch=lx24-x86 #$ -o /export/home/nico/cyp3a4/stdout #$ -e /export/home/nico/cyp3a4/stderr #$ -i /export/home/nico/cyp3a4/stdin qsub md-job.sh # # needs in # # # # $NSLOTS the number of tasks to be used $TMPDIR/machines a valid machine file to be passed to mpirun echo "Got $NSLOTS slots." /usr/bin/mpirun -np $NSLOTS -machinefile $TMPDIR/machines /export/home/nico/gromacs/bin/mdrun_mpi -s /export/home/nico/cyp3a4/1TQN -md.tpr -o /export/home/nico/cyp3a4/1TQN -md.trr -c /export/home/nico/cyp3a4/1TQN -after-md.gro -np 16 -v File Cyp34a (GPU) Convert File Into .topology & .gro pdb2gmx -f 1TQN.pdb -p 1TQN.top -o 1TQN.gro Periodic Boundary Condition editconf -f 1TQN.gro -o 1TQN.gro -d 1.0 Adding solvent into the molecule genbox -cp 1TQN.gro -cs spc216.gro -p 1TQN.top –o 1TQN-solvate.pdb Energy Minimization grompp -f md.mdp -c 1TQN.gro -p 1TQN.top -o 1TQN-md.tpr Production Simulation mdrun-openmm -v -deffnm 1TQN-md File Curcumin topol.tpr is ready to be executed with 10 processor grompp -np 10 -f md.mdp -c lox_pr.gro -p model.top -o topol.tpr dt x nsteps = …pikosecond 0.002 x 100000 = 200 pikosecond Md-job.sh #!/bin/sh # Curcumin #$ -N gromacs #$ -cwd # Jumlah prosesor #$ -pe mpich 10 #$ -l arch=lx24-x86 #$ -o /export/home/ari/simulasi/curcumin10/stdout #$ -e /export/home/ari/simulasi/curcumin10/stderr #$ -i /export/home/ari/simulasi/curcumin10/stdin # # needs in # # # # $NSLOTS the number of tasks to be used $TMPDIR/machines a valid machine file to be passed to mpirun echo "Got $NSLOTS slots." /usr/bin/mpirun -np $NSLOTS -machinefile $TMPDIR/machines /export/home/nico/gromacs/bin/mdrun_mpi -s /export/home/ari/simulasi/curcumin12/topol.tpr -o /export/home/ari/simulasi/curcumin12/curcumin12.trr -c /export/home/ari/simulasi/curcumin12/lox_pr.gro -np 10 –v qsub md-job.sh File Curcumin (GPU) grompp -f md.mdp -c lox_pr.gro -p model.top -o curcumin.tpr dt x nsteps = …pikosecond 0.002 x 100000 = 200 pikosecond Production Simulation mdrun-openmm -v -deffnm curcumin Performance Result File : Curcumin Performance Time Timesteps Single Processsor 24h:01M 200 ps GPU (GTS 250) 17h:01M 200 ps GPU (GTS 250) 9h:24m 100 ps File : CYP3A4 Performance Time Timesteps Single Processsor 22h :32 M 200 ps GPU (GTS 250) 14h : 23M 200 ps GPU (GTS 250) 7h : 45 M 100 ps