There is literature on why this is the right thing to do for such models.

Just mention that the new thing is that the cells now additionally express CCR5.

Change the color scheme.

Use post-fusion conformation and Andrew Ward CD4 bound.

This would be:

1. Additional structures, either figure supplements or additional panel in Figure 7 showing shifts qualitatively.

2. Also be repeating proximity to shifted sites analysis in Figure 7C using each of the other structures and maybe one where it's minimum distance across structures. We might want to calculate some metric of how well these different distances perform, but first let's just plot them. We might want to repeat with other panels as well.

3. Right now, distances are computed from PDB using dms_tools2.protstruct.distMatrix function in the notebook. If you generate clean versions of the relevant PDBs for the other structures similar to the 5FYL_Env_trimer_rmTER.pdb in ./data/, I can also re-run that code on additional structures easily.

We can explain a bit, but basically we should put wildtype amino acids for Figure 6C.

We also need to clearly relate to Figure 7 - figure supplement 1.

Jesse will first look to see if wildtype labeling can easily be done with code. If not, we'll hand-label them after we make sure everything else is stable in figure.

Figure 1A describes a "12 hour infection". However, we actually did a 3 hour infection, spun cells down and resuspended in fresh media, then harvest cDNA 12 hour post infection. We could change the text to "infect, harvest viral cDNA 12 HPI". Or maybe these details don't matter for the figure as long as it is in the text.

Probably best addressed by zooming in on a few sites in detail and talking about their interactions?

Maybe best to do this after we've done major revision 2.

Just add some explanation.

There has been a phydms bug fix since the original submission. We should re-run with newer version.

Can explain textually why we did this, and why it is equivalent to re-scaling log enrichment ratios.

I think the problem was that it was using a newer version than was on PyPI, but now it should work. At the end, we'll make sure it works with current version and update version reference in paper.

We should do this last.

@jbloom: I also found a problem with the wording in Figure 9. I wrote the the color scale ranges between "−6 < log10 𝑄 < 6". This is true for the directional log10 𝑄 values that account for omega being greater than or less than one, but not the raw log10 𝑄 values.

This is just a reminder for one of us to change this before re-submitting.

@jbloom: I was looking at some of the edits you made and found a problem with the wording in the caption for Figure 7D. It says "crystallographically" resolved sites a few times, but the 5VN3 structure was determined using cryoEM. So, a more accurate wording would probably be "structurally" resolved, or something like that.

I raised this issue to as a reminder for one of us to address this before submitting. I'm happy to do it now if it wouldn't conflict with your revisions.

I think other schemes work even worse, but we could make heatmap / contour plot and either show as supplement (if it looks kind of OK) or just in reviewer response (if it looks terrible).

We need to answer the questions e-mailed to Jesse for the eLife digest.

There is definitely a lot on all proteins.

For Env, we add Arup PNAS paper and say something in response about how it was published after ours was submitted.

Are there other relevant papers? Maybe we should carefully read the recent PNAS Arup paper and see if they cite others that we should cite. We can be generous here about citing a lot of stuff.

Make this clearer.

This is something that needs to be clearly articulated.

We should explain rationale for region we mutated along with some references about how SP and CT affect expression.

Somewhere either in Results or Discussion, we should add a sentence emphasizing how we are measuring overall effects on viral growth which are combination of expression level, stability, and per-molecule function.

Similar to minor comment 3.

@Haddox: Could you make a structure or structures that show the sites of diversifying and purifying selection? We would want these to go in Figure 6. I have left some notes in Figure 6 about what might go in there, and there is a data file referred to in the figure legend that has the information to plot.

You can just make the figure (along with the script that generates it) in an appropriately named subdirectory of ./paper/figures/