Mac Pro vs. Windows 10 for MinION and Downstream Bioinformatics

[Oetam]

It looks like we may be purchasing a new computer to run our newest sequencer (Oxford Nanopore MinION). We are looking at a desktop machine (not a cluster) that we can use both for running the MinION and for the downstream bioinformatics (e.g. samtools, bwa, canu, nanopolish). I am a wet labbie who has been given this task, so I am looking for an easy-to-use solution that will not break the bank. I would be very interested to hear any opinions regarding our options:

1. Apple Mac Pro ($$$ and less powerful than the rest)
2. Custom workstation running Windows 10 (Does the new Ubuntu bash work well enough? I have seen a lot of issues with Cygwin in my Googling)
3. Custom workstation running Ubuntu 14.0.4 (The sequencer software doesn’t seem to work well in Ubuntu)
4. Custom workstation with Windows 10/Ubuntu 14.0.4 dual boot (Was told this is not stable)
5. Custom workstation running Windows 10 and Ubuntu 14.0.4 virtually (Bioinformatics performance reduced?)
6. Custom workstation running Windows 10 and booting Ubuntu off of a USB (Bioinformatics performance reduced?)
7. Other options?

The Mac Pro seems most likely to work straight out of the box and give us the least headaches, but it is also much more expensive than comparable Windows/Linux boxes. It seems that the MinION software is developed for and works best under Windows, and from what I can tell, there have been issues with Linux. Has anyone used the new Bash shell integrated into Windows 10 to get some of the most standard bioinformatics programs running? Alternatively, does anyone have experience with how virtualization impacts bioinformatics performance?

Any input would be great. Thanks!

[wdecoster]

Based on the community forum there are occasionally temporary issues with both Windows, Mac and Linux MinKNOW. So I wouldn't make a decision based on that argument. The most "flexible" solution is probably the dual boot. I'm not sure about windows 10 in dual boot, though. I've used it with earlier Windows versions.

If you feel comfortable enough using Ubuntu and/or are prepared for learning new things (command line) I would suggest to give that a try. It will make a lot of downstream bioinformatical steps far easier than windows/cygwin/virtual box solutions.

[Oetam]

Thanks for your response. I appreciate it. I am not very comfortable with the Linux command line outside of simple commands/scripts, and although I will have to implement the downstream bioinformatics programs, I was trying to figure out what would be the simplest solution for me regarding MinKNOW/Metrichor (command-line only for Ubuntu) and the analysis pipeline. It is probably the Mac assuming there are no issues with the assembly/mapping programs running on the device, but based on your advice and that of others, I might go for Linux as we can get a cheaper and more powerful machine.

Data integrity is important for us, so I think we might try to avoid the dual boot.

[macarima]

I'll choose '3. Custom workstation running Ubuntu'.
I had 4 workstations with ubuntu 14.04 and 16.04.
I don't have any problems to assemble genomes with various assemblers like AllPaths-LG, Canu, Ray, Platanus and so on. I also have a mac pro 2013 late. But I don't use it for assembly, because of limited RAM and less powerful CPU.