Sally's changes to Figure 1 (sc2ts schematic)

[hyanwong]

Sally made 3 comments on @szhan's sc2ts schematic (which used to be figure 7, now moved to fig 1)

1. The node distinctions are not consistent with the subgraph plots (e.g. Fig 6). For example, we could have non-sample nodes as smaller, and recombination nodes as black (not blue). I'm not sure we need to make them consistent (and @jeromekelleher agrees), but I guess if it is easy to do so, it wouldn't hurt.
2. In figures D and E, the "newly added sample" nodes should presumably be at a different time (the most recent time), and none of the previously added sample nodes should be at the bottom of the tree
3. Figure D is logically problematic, because the middle node should have been added to the left hand node.

Along with point 3., there is a problem because we need to indicate that the two mutations are actually the same mutation. We need some way to indicate that. One possibility is to use a different symbol?

[hyanwong]

After discussion, for point 3, rather than using a different symbol, perhaps a label on subfigures C and D pointing to "identical mutations" would be easiest?

[jeromekelleher]

@szhan - do you think we'll get these in today?

[szhan]

WIP.

1. I have updated the Legend. I've added the symbols for recombination nodes and inserted nodes that Yan uses in the other figures.
2. In panels D and E, the older sample nodes are moved up, situated higher than the newly added sample nodes.

The positions of the nodes in panels D and E needs some tweaking. Also, I still have to add labels next to the nodes.

[szhan]

I also removed the texts "Mutation-collapsing node" and "Reversion-pushed node" in panels D and E. I think that it is clear enough without them.

[hyanwong]

Thanks Shing. I reckon you can probably move the older sample nodes downwards a fraction more - no need for them to be halfway between the newly added samples and the root?

[szhan]

I think these labels make it more clear?

[hyanwong]

They do help, yes. But I wonder if it might still help to indicate somehow that the two mutations are the same? Maybe it's a bit clumsy though?

[szhan]

Or just highlight them?

[szhan]

This is WIP of panel E.

[szhan]

I still somewhat prefer having the middle sample node half way.

[szhan]

[szhan]

Maybe the text highlighting in panel D is not obvious enough. Not when the full figure is displayed and panel D is not zoomed in.

[hyanwong]

I think the mutation labels are quite small to read - they might be overlooked. But it is nice to have them, and they are informative.

[szhan]

I don't see harm in using more colours, since we have only used three non-greyscale colours so far.

[hyanwong]

Good point. I think colours on the mutation bars themselves would be confusing, but colours on the labels wrks well, I think.

[hyanwong]

Can you make the mutation label font about 10% bigger, or does that cause problems?

[szhan]

Actually, I think I'mma change the colours for all the symbols. We could, for example, make the text labels in panel C to pop by using red. Then, use a different colour for mutations, say, blue. Parent 1 could have a different colour, say, orange.

[hyanwong]

Hmm, I'd be careful. Too many colours can be confusing. It's good to use them to highlight very specific things.

[hyanwong]

Personally, I think what you have is perfect, colour-wise (maybe slightly darker blue for A100G)

[szhan]

How about this?

[jeromekelleher]

LGTM

[szhan]

Checking once more and then PR.

[szhan]

Missed that one mutation in on the right local tree. Darn.

[hyanwong]

Mentioned to Shing on Slack - currently it's also missing labels to show mutations are identical in subfig C...

[szhan]

[szhan]

Which set of colours is better?

[hyanwong]

I like the right hand (orange text).

I think, however, that we should be consistent and also have the same colour text in 1C.

[szhan]

[szhan]

Latest version

[hyanwong]

Looks fine. Is there any way to avoid the dotted arrows in C. by placing the label twice, like it is in the panel below that (I realise there might not be room, though). Something like:

The disappearance of the second orange label makes it nicely clear what is going on.

Also, I guess I could change the mutation labels in the subplots to the blue you are using for mutation bars? This is what that would look like. It would, I guess, be nice to use bars instead of numbers for the mutations, to match your schematic, but I think that's too much work for the preprint.

[jeromekelleher]

Are we happy with the latest version just merged?

[szhan]

I kind of like the blue mutation labels. Maybe let's go with that for now, since it is quickest?

[szhan]

Removed the dotted arrows.

[hyanwong]

Look great to me. Do we want to add something to the legend for describe "G7A", e.g. instead of "Mutation", we could say "Mutation (may be labelled e.g. G7A for inherited state G, position 7, derived state A)". Or a shorter label if you can think of one?

[szhan]

I see labels like "G7A" often in the literature, so maybe our target audience is comfortable with it.

Also, in the legends, recombination node, inserted node, and sample nodes are listed in different orders. Should we sync them too? If so, maybe stick with that order?

Maybe I should lowercase the text in the legend as well.

[hyanwong]

I see labels like "G7A" often in the literature, so maybe our target audience is comfortable with it.

Yes, or we could commend in the caption instead. A shorter legend description would be "Mutation (may be labelled with state change and position)"

Also, in the legends, recombination node, inserted node, and sample nodes are listed in different orders. Should we sync them too? If so, maybe stick with that order?

Good point. I don't think it's important, but I guess sample nodes are the first ones encountered, which is why I put them first. But we could go for alphabetical order:

inserted recombination <- this is a type of inserted node, so might be useful to put after the inserted node sample node sample node (newly added)

Maybe I should lowercase the text in the legend as well.

I'm happy with either. Which looks best to you?

[jeromekelleher]

Can we close since #175?

[hyanwong]

I think close and we can open up a different issue for the minor tweaks discussed above, as they are very minor indeed.