Computational Choices

Structure solution in protein crystallography no longer requires specialized or exceptionally powerful computing equipment. Most modern desktop , laptop , and even tablet computers have sufficient power to carry out the necessary tasks. Some personal thoughts about hardware and operating system choices are given here.

Operating System

Protein crystallography software is mostly developed and compiled within a Linux environment. While most software has been ported to both Windows and Apple OS operating systems, in some cases not all of the original features survive the transition. The choice of operating system is purely personal, but there are advantages and disadvantages of each.

Linux

The Linux operating system is free, making it possible to build inexpensive dedicated workstations for student or lab work. Linux works well with modest CPU and memory. Driver support for hardware in Linux can be challenging at times, especially with the latest, greatest releases, but has dramatically improved with current distributions. I recommend using a long-term support (LTS) version of Ubuntu. The most current version is 18.04 LTS. LTS releases are less buggy, and if you are not using the most recent version, bug remediation and hardware driver support is mature. Installing software and updating the operating system is very fast, and does not normally require system restarts in Linux. There is a learning curve for maintaining a Linux system, but it is, in my opinion, well worth it.

Windows

Windows is probably the most popular and familiar operating system. Most crystallography software has been ported to Windows, but not always gracefully. Windows systems will generally require faster processors and/or more memory than the Linux equivalents to achieve equal performance. On the plus side, hardware support is generally excellent.

MacOS

Apple systems are in some ways the best of both worlds, sporting familiar, well-supported operating system, plus a Unix shell underneath that is fairly compatible with protein crystallography software developed in Linux. The principal disadvantage of using an MacOS system is that it is the most expensive option, and not all users will be familiar with or own an MacOS-based computer system.

Dual Boot Machines

Many users employ dual boot machines, with both a Linux and Windows OS installed. It takes sufficient hard drive space for two operating systems, plus some technical expertise to accomplish, but is a viable option.

Virtual Machines

The use of a virtual machine for protein crystallography is not universally recommended, as much of the software for conducting protein crystallography structure solution is computationally and graphically intensive, where hardware acceleration is important. While it is possible to run crystallographic software in a virtual machine, it can be very slow or laggy due to software emulation of hardware functions. However, virtualization is improving. (See Chromebooks, below.)

Chromebooks

Modern (2019 or later) Chromebooks can be partitioned to run a suprisingly good virtualized version of Debian. Pymol (apt-get install pymol), CCP4i, and Coot run quite well on machines with an 11th generation Intel i3 or better processor with 8 GB of RAM, and are perfectly suitable for basic work. This is an intriguing and cost-effective option for student work.

Hardware

Protein crystallographic software will run on desktop computers, laptops, and tablet computers. However, some of these platforms may be limited in hardware options. Having said that, I have run protein crystallography software suites on a Microsoft Surface Pro 6, laptop computers, typical office desktop computers, and home-built Linux mini-towers. Making good hardware choices will make your structure solution sessions more pleasant.

Monitor

Spring for a large, high resolution monitor. Protein crystallography is highly graphical! For a desktop system a 24-27 inch HD or 4K monitor is ideal. For laptop or tablet computers the highest resolution possible, preferably 4K, is best.

Mouse

Crystallography software requires a three-button mouse with a scroll wheel. While it is possible to emulate the middle mouse button and the scroll wheel in two-button or button-less mice, this will not be a pleasant experience to use. Invest in a comfortable 3-button laser mouse.

CPU

Most modern CPUs will be quite capable of running protein crystallography software. A quad-core CPU or better is recommended, as some crystallography software can use more than one core, or you will be able to run multiple processing jobs at once without a performance lag. However, many programs are not compiled for multi-CPU use, so each individual job will run on only one core at a time. Therefore, it is an advantage to choose a CPU with a fast as possible clock speed. That is, a quad core CPU with a faster clock speed will perform better than a six-core processor with a slower clock speed.

Memory

The amount of memory (RAM) required will vary by operating system. In Linux as little as 4 Gbyte is often sufficient , but a minimum of 6-8 Gbyte would be better. For a Windows or MacOS system, 8-16 Gbyte is probably a better minimum requirement.

GPU

For protein crystallography computing systems, Nvidia rules. Don't even think about other GPUs or graphic cards. Put simply, the vast majority of crystallographic software is well supported for Nvidia hardware, and if you want to implement stereo viewing, NVidia is the most well-supported graphics system. Integrated Intel graphics and Radeon graphics hardware are powerful, but are frequently plagued with bugs in the protein crystallography software ecosystem. Seek out a laptop with an integrated Nvidia graphics system, or invest in an affordable Nvidia graphics card for desktop systems. A GTX-1050 graphics card costs less than $200.