Linux or Windows for 454 assembly of bacterial genomes?

[mbseq]

I know this has been covered before but I also know that the answer is often "it depends". So here is a brief description of what I am trying to do.

I am new to NGS and working on sequencing small (1 Mb) bacterial genomes (http://seqanswers.com/forums/showthread.php?t=24214). I have sequence data for 3 isolates and it appears that I have many gaps (about 80) in each of the assemblies as a result of repeats. So for finishing I have been doing de novo assembly with Newbler, moving contigs with Mauve, and using Ugene for alignments and primer design. This is a lot of work and will get expensive .

Thus far I have been restricted to using Windows for analysis beyond the Newbler assembly. I have been given the go ahead to purchase a computer to install Linux and give me more options for data analysis. But without the experience I am not sure it will necessarily help me out.

Does it seem worthwhile to go ahead with purchase? Will I be able to improve my assemblies and reduce the amount of bench work needed for finishing? Convince me! What kind of computing requirements am I looking at?

[jimmybee]

Why don't you run a virtual machine of a linux operating system (say Ubuntu or something similar)? Its great to get your feet wet in the linux side of things, and theres few issues if you have problems or break it. If your windows machine has enough resources you will be able to run windows and linux side-by-side. An 1Mb genome is quite small so you don't have to have massive resources.

If, after playing with linux for a while, you feel like you need more resources then you can go ahead with your purchase of a new machine

[JackieBadger]

I agree.. We use a high end windows based machine running ubuntu on a virtual machine. You get the best of both. The benefits you gain from a unix pipeline is adaptability i.e. there are lots of programs available for you to mix n match. Ubuntu has a GUI and with a little know how most NGS programs can be run with relative ease....the added benefit is that command line programs seem much more efficient at dealing with large data sets.

[xied75]

Another suggestion even simpler is you can go AWS or Azure to get a trail account then in 5 mins you can start a new VM running Ubuntu, in this you can concentrate on getting familiar with Linux and those command line tools. If you did something wrong, delete the VM and start again.

Best,

dong