Runtime assessments/using EC2 to benchmark

[benedictpaten]

To write this up we need to record runtimes for how long it takes (wall and processing time) to process a minion dataset, both for marginAlign and marginCaller.

I think it would be good to do this on EC2, and to use docker to create a container that would run on EC2 as a push button process - that way an outside group without compute could use a beefy EC2 instance to process minion datasets with our tool set - perhaps they could pipe direct from the Metrichor process to our processing pipeline. Let's discuss this in person.

[nadakubikova]

Hi guys, Where things stand with using EC2 to benchmark for those who don't have the compute but need to process MinION data sets :D. I am finding that the run times are okay for me when aligning against a single chromosome but not against whole genome.

[mitenjain]

Hello,

We have implemented a version that uses toil and works on EC2. This was used for the nanopore human genome paper. It should be still functional and needs some code tweaking. We can make that available in a couple weeks time and will add a link here for the code.

-Miten