[User story] HiMap testing

[richardjgowers]

In broad terms, what are you trying to do?

@pitmanme has created HiMap -> https://github.com/MobleyLab/HiMap (this repo is representative, but also lags behind a revised version of the software coming soon)

This has a new way of proposing/optimising perturbation networks (not new atom mappings, but which edges to choose).

This should hopefully fit within the existing planned API points.

How do you believe using this project would help you to do this?

• It would be good to run all vs all networks (or all vs all within given clusters) to generate exhaustive data to check the optimal solutions suggested by HiMap.
• Back to back testing of Lomap vs HiMap (or any other network proposal tool) on different benchmark series
• Trying variations on proposed networks to fine tune the algorithm

What problems do you anticipate with using this project to achieve the above?

The possible protocol space (number of protocols that one could try networks with) to try on F@H is limited, so conclusions might be limited by that.

[davidlmobley]

If computing resources prevent it, one could also run calculations which are inclusive of (edges suggested by LOMAP) plus (edges suggested by HiMAP) so one can compare the full set, the individual subsets, etc.

Note that HiMAP will typically suggest more edges than LOMAP, especially as the graph size grows.