search

choderalab / ensembler-manuscripts

Manuscript for Ensembler v1
0 stars 3 forks source link

Please take a final edits pass! #51

Closed jchodera closed 8 years ago

jchodera commented 8 years ago

We're almost ready to resubmit!

Please take a final pass through both the manuscript and the referee response letter @sonyahanson @kyleabeauchamp @danielparton @pgrinaway

danielparton commented 8 years ago

All looks good to me! I'm making a new Ensembler release now, so will update the referenced Ensembler version in a moment.

danielparton commented 8 years ago

Ok, I've made a new Ensembler release and updated the version number referenced in the manuscript. I also updated the text very slightly to reflect the following changes:

jchodera commented 8 years ago

Thanks!

We still need @sonyahanson @kyleabeauchamp @pgrinaway to check in here.

sonyahanson commented 8 years ago

On it.

pgrinaway commented 8 years ago

On it as well.

kyleabeauchamp commented 8 years ago

Will get to this tonight or tomorrow morning. On Oct 16, 2015 8:33 AM, "Patrick Grinaway" notifications@github.com wrote:

On it as well.

— Reply to this email directly or view it on GitHub https://github.com/choderalab/ensembler-manuscripts/issues/51#issuecomment-148747950 .

sonyahanson commented 8 years ago

Overall looks good! New stuff incorporated nicely. No major comments.

A few minor comments on the response letter:

A few minor comments on the manuscript:

jchodera commented 8 years ago

Thanks for the comments! Any chance you can fix those things and check in changes?

The red/orange bit might have been parts where I was asking Danny to proofread. On Oct 16, 2015 1:36 PM, "sonyahanson" notifications@github.com wrote:

Overall looks good! New stuff incorporated nicely. No major comments.

A few minor comments on the response letter:

  • Since this is so long, might be nice to have page numbers?
  • The fonts can be a little inconsistent. Specifically on p.2 font briefly shifts from callibri to arial.
  • Should we mention that the current version of ensembler does somewhat address the problems of loop modeling (through rosetta) and protonation states (through the manual overrides yaml file) in the response to review 2?
  • In paragraph citing Daggett Dynameome work responding to review 2: 'superamily' - 'superfamily'
  • Is the yellow and red bit in responding to reviewer 2 supposed to be in the final version? This is a bit jarring if so.

A few minor comments on the manuscript:

  • There's a weird space on p. 3 between lines 200 and 201.
  • Two references on p. 12 showed up as question marks - lines 848 and 852.

— Reply to this email directly or view it on GitHub https://github.com/choderalab/ensembler-manuscripts/issues/51#issuecomment-148780667 .

kyleabeauchamp commented 8 years ago

Overall looks good. Comments below:

  1. Page 4, line 214. Our text seems to suggest that common UniProt protein family membership guarantees homology. I'm not 100% sure whether that is true. I imagine that UniProt membership suggests homology, but doesn't guarantee it. See also: http://www.uniprot.org/help/family_membership
  2. Page 12: box 2. We might consider adding a sentence to the caption explaining that users actually need to fill in their own Modeller license key in the string. Some nontrivial fraction of users will copy paste the box and then send you correspondence when it fails...
  3. P12 Lines 848, 853. There are undefined citations with "?".
danielparton commented 8 years ago

@jchodera I'm trying to work out which missing citations are meant to go on p.12. This is the text:

An alternative approach could be to re-refine complete-chain template structures to the experimentally-derived electron density or scattering data deposited in the RCSB using methods capable of exploiting the scattering data and crystallographic symmetry~\cite{schnieders:biophys-j:2010:x-ray-refinement}. Even if definitive placement of these unresolved regions is impossible, plausible locations constrained by weak scattering data and strong steric exclusion of crystallographic neighbors may provide a great deal of useful information, especially when combined with forcefield priors~\cite{schnieders:jctc:2011:space-group-pme}.

I think the citations are:

  1. Fenn TD, Schnieders MJ, Brunger AT, Pande VS. Polarizable Atomic Multipole X-Ray Refinement: Hydration Geometry and Application to Macromolecules. Biophys J. 2010;98: 2984–2992. doi:10.1016/j.bpj.2010.02.057
  2. Schnieders MJ, Fenn TD, Pande VS. Polarizable Atomic Multipole X-Ray Refinement: Particle Mesh Ewald Electrostatics for Macromolecular Crystals. J Chem Theory Comput. 2011;7: 1141–1156. doi:10.1021/ct100506d

I'm adding them to the bibtex file now, but let me know if these are wrong.

danielparton commented 8 years ago

@kyleabeauchamp regarding p. 4 line 212 That line "guaranteeing homology..." was actually the reviewer's suggested rephrasing of my original phrase "guaranteeing some degree of homology..." because he thought the phrase "some degree of homology" was meaningless. I agree with you both, and am now changing the line to "establishing homology...", which hopefully is satisfactory for all parties :)

danielparton commented 8 years ago

@jchodera @sonyahanson @kyleabeauchamp thanks for the comments. I've implemented all suggestions in both response letter and manuscript. Except for fixing the weird white space - got to leave something for the journal copy editors to do...

Sonya, regarding the discussion of loop modeling and protonation states in the response, I just removed "loop modeling" from the following phrase: "Similarly, a number of other limitations exist (loop modeling, cofactors, ligands, structural ions, protonation states, etc.) in this current version of the software" Protonation states can indeed be defined manually now with the yaml configuration file, but for the response letter I think it would only be worthwhile to mention a more advanced automatic assignment algorithm, which we haven't implemented (yet).

pgrinaway commented 8 years ago

Overall, looking good. Comments shortly.

pgrinaway commented 8 years ago

Some minor notes:

Does it matter that we say low sequence identity models (L753 p11) can push the simulation to new relevant areas, but then later (L947, p13) state that the very low sequence identity models are likely to result in simulations of inaccessible regions of configuration space (in terms of efficiency)?

With regard to the response letter, I am wondering what the reviewer actually means in suggesting "monte carlo based MD method" Obviously, monte carlo MD is HMC, but I wonder if the author actually meant to suggest some sort of non-MD MCMC scheme, and whether it would save some headache to address why MD vs some Metropolis Monte Carlo with no MD.

jchodera commented 8 years ago

I think the citations are:

Those are correct! Sorry for not having provided those!

danielparton commented 8 years ago

@pgrinaway thanks for the comments.

Regarding first point - the idea is that low sequence identity models are less likely to be thermally accessible, but by the same token are likely to include models which represent a wider range of thermally accessible conformations than those covered by high sequence identity models. Weeding out the inaccessible conformations can be done at the MSM construction stage.

So I think the text can be left as is, except perhaps we should take out the word "highly" from the second passage: "many models---especially those derived from very low sequence identity templates---are likely to be highly unrepresentative of conformations populated at equilibrium by the target protein."

Regarding the second point - if the reviewer has used the phrase "monte carlo based MD method" wrongly, then he can clarify in his next review if it is really important to him.

@jchodera I think we are ready to wrap this up now?

jchodera commented 8 years ago

Thanks everyone for checking in! I'm taking a final pass and will get this resubmitted today.