General usage

As described here, data analysis in MDI tool suites is divided into two stages called pipelines, which perform high-performance computing on Linux servers, and apps, which support interactive data visualization in R Shiny.

At present ont-mdi-tools only offers pipelines.

Methods and help for calling MDI pipelines

Please see the detailed documentation and MDI command help:

mdi --help

for information about the ways you can execute an MDI pipeline on your server. Briefly, you can run a pipeline action as a program from the command line, e.g.:

mdi <pipeline> <action> [options] # e.g., mdi ont basecall

However, rather than specifying options at the command line, we recommend creating a job configuration file and then either calling it directly:

mdi <pipeline> <data.yml> # e.g., mdi ont mydata.yml

or, better yet, submitting it to the job scheduler on your HPC cluster:

mdi submit <data.yml> # e.g., mdi submit mydata.yml

Help for assembling job configuration files

Complete instructions for constructing MDI job files are found here - there are many additional helpful features.

• https://midataint.github.io/mdi/docs/job_config_files.html

The following command will print a template you can use to quickly construct a new job file from scratch.

mdi <pipeline> template --help
mdi <pipeline> template > mydata.yml # e.g., mdi ont template
nano mydata.yml

Finally, the following commands will show help for a pipeline or one of its actions to understand how options are organized and what they do:

mdi <pipeline> --help          # e.g., mdi ont --help
mdi <pipeline> <action> --help # e.g., mdi ont basecall --help

Using ont-mdi-tools code outside of our pipelines

You may use whatever code you’d like from this repository in your own pipelines subject to the MIT license.

In particular, shared modules repack and basecall offer standalone scripts for performing IO-intensive actions on POD5 files on shared HPC resource servers.