4. At least 2 NVIDIA graphics processing units (GPU), each with
at least 6 GB of memory. We used a computer with 4 NVIDIA
GTX1080s, each with 8 GB of memory (see Note 1).
5. If your data is stored on a central storage system, a local solidstate scratch disk (SSD) is recommended. We used a
400 GB SSD.
2.2 Software
Requirements
1. A Linux-based operating system. We used Scientific Linux 7.
2. CUDA. We used version 9.0. Download from https://
developer.nvidia.com/cuda-zone.
3. MPI. Any MPI flavor may work. Not only the runtime
(mpirun) executable, but also the header files are required to
compile RELION. We used OpenMPI version 1.5.4. Download from https://www.open-mpi.org.
Fig. 3 FSC plot generated by the post-processing job-type of RELION for the helicase body. See Note 16 for
details
Fig. 4 (a) The contributions of all principal motions to the variance. (b) The third and eighth map along the
second principal motion, which is highlighted in red in panel a
148
Takanori Nakane and Sjors H. W. Scheres
at least 6 GB of memory. We used a computer with 4 NVIDIA
GTX1080s, each with 8 GB of memory (see Note 1).
5. If your data is stored on a central storage system, a local solidstate scratch disk (SSD) is recommended. We used a
400 GB SSD.
2.2 Software
Requirements
1. A Linux-based operating system. We used Scientific Linux 7.
2. CUDA. We used version 9.0. Download from https://
developer.nvidia.com/cuda-zone.
3. MPI. Any MPI flavor may work. Not only the runtime
(mpirun) executable, but also the header files are required to
compile RELION. We used OpenMPI version 1.5.4. Download from https://www.open-mpi.org.
Fig. 3 FSC plot generated by the post-processing job-type of RELION for the helicase body. See Note 16 for
details
Fig. 4 (a) The contributions of all principal motions to the variance. (b) The third and eighth map along the
second principal motion, which is highlighted in red in panel a
148
Takanori Nakane and Sjors H. W. Scheres