check dms-viz visualizations

[jbloom]

@arjunaditham, I have added dms-viz JSONs as part of the pipeline.

These can be viewed at a section near the bottom of the docs.

If you provide the raw links to the JSONs (eg here) to dms-viz then you get interactive views, as in here.

Can you go through all six JSONs and make sure they seem correct to you in terms of putting the DMS correctly on the structures? I wanted to make these so I could look at them alongside the paper as I work on it.

Note that the one with all the antibodies (eg, here) will be better once @WillHannon-MCB addresses this issue, but it is still useful now I think.

Also, it should be pretty easy to configure more of these using the YAML configuration here, so either add or flag for me if you want more structures added.

Also, @Caleb-Carr, I wanted to flag this for you as well as I know you are doing a lot of Lassa antibodies. Right now I added this as a custom rule with:

• configuration here
• custom rule here
• notebook implementing the rule here

But I would like to eventually add to the pipeline. As time permits (it is not high priority), would you be interested in also trying to integrate a similar rule into your pipeline, and then somehow tracking (via this or some other GitHub issue) if it just works and if not we can try to fix in a modular way. This could be a step towards making sure it works robustly to eventually putting into dms-vep-pipeline-3.

[arjunaditham]

@jbloom I took a look at these and these are correct. The only point of feedback I have is that we typically have a minimum filter of times_seen of 2. I did not currently see that implemented (though I might have missed that).

I think it would be possible to implement in YAML either as a limit or a filter. (I can also try doing this myself if you'd like). This is a very modest change, but it would be consistent with our prior decisions and how we typically examine the heatmap and identify escape mutants with confidence.

[jbloom]

@arjunaditham The times_seen filter is already implemented on mutations shown in the summary plots. That's why I was suggesting that @Caleb-Carr use those CSVs rather than the custom rules to do this filtering (see here).

So the idea in the summary plots and now these dms-viz ones is that we have already filtered for high confidence measurements with times_seen, and are showing just those high-confidence values with biologically meaningful filters (eg, cell entry).