Review Reports

Round 1

Reviewer 1 Report

The authors did a good job introducing the subject, in a very complete and concise manner. However, when reading the results and the analysis done, this sounded more like preliminary data than actual irrefutable results. Most of the data requires further work and it seems they still have a long way to go. I recommend they re-submit the paper with additional tests, for instance by adding data associated to the 4 points highlighted in the conclusions (essential in my point of view), that in my opinion will improve the quality and the general importance and impact of the work.

The English writting also requires more attention, particularly the abstract and introduction.

Author Response

We have included further analysis of the data, in particular multiple regression that allowed the bacterial burden data to be treated as a continuous variable.  This highlights and explains some trends we saw with PCA.  In addressing the 4 points discussed in the conclusion:

1) find ways to improve the workflow to save compute resources;

we included a diagram of the workflow and discussed how using coarsegraining would save compute resources

2) streamline structure representation with residue coarse grain;

 

3) docking simulations to show conformation adapted while adsorbed with additional combination structures that can augment the dataset; and

This is important if we use an adsorbed protein, however, we are proposing surface conjugated protein, so at present this is not necessary 

4) optimize our descriptor set to allow informative predictions of biocompatibility and biofouling.

the added multiple regression results have given some informative predictions

 

Author Response File: Author Response.docx

Reviewer 2 Report

Dear authors

The manuscript is well-written and the experiments are described in details.

However, I miss some applications, I mean, are they only theoretical calculations? How do you plan to apply to biomaterials?

Do you plan to reply with other molecules?

 

Author Response

Thank you for the kind statements.

are they only theoretical calculations?

Prediction of molecular descriptors is using structural information is done fairly often (ref 17).  In our data for lysozyme, in tables 1 & 2 the dipole moment and hydrodynamic radius are the same order of magnitude as measured using dielectric spetroscopy.  For hydrodyamic radius in pure water solvent, the calculated value is within 10% of the upper error bar of the experimental value.

How do you plan to apply to biomaterials?

We expanded this in the revised conclusion and included a an additional figures for explanation.  Protein coatings for metal surfaces, have a variety of applications aside from coating medical devices such as Stents and Joint replacements, these include biomaterial testing using SPR.

 

Do you plan to reply with other molecules?

We used GelMA and some varieties of troponin, we also included lysozyme because it is a benchmark for biofouling.  While it's true that the proteins are larger than the peptides used as our target data for prediction, the differing descriptor values such as rg should account for the larger size.  The target data peptides are also dimerized giving the training set added variety.  GelMA is a good coating protein for biocompatibility because of the density of RGD sites   as seen in revised figure 2.  So that's why we used it instead of a polymer such as PEG, ref 17 also mentioned problems of PEG buildup in vivo after long-term use.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors did a good job responding to the reviewers requests. There are still some small English mistakes that require further attention but aside from that the information is clear and the data is well discussed.