Documentation and Examples Update

[bannanc]

This pull request is addressing concerns from @davidlmobley and @jmaat largely voiced over slack, but also in issue #25. Below is the big picture of what will be changed here, mostly like minute details will still be discussed on Slack.

Examples

• [X] Update Single Molecules example to include more explanation at the top and more details about what to look for in results throughout
• [ ] Update SMIRKS from Molecules example to have a better introduction/big picture at the top and explain in detail what is going to happen before each code box. Will also fix the typos Jessica pointed out (if they don't get rewritten completely)

Documentation

• [ ] Change to introduction to focus more on big picture, why this tool is necessary, replacing atom types, automating chemical perception generation for force fields and why that is necessary. The main point being sell this tool here, not the whole detailed history of smarty/smirky/etc.
• [ ] This section needs to have a more clear definition of what chemical perception is, how it is currently generated and how chemper will improve things
• [ ] Move move of the current intro into a "history/off connection" section, I'm still trying to come up with the best word for this area, maybe History is sufficient, or for A Lot More Details
• [ ] Fix minor typos
• [ ] Make sure Future wraps back to the big picture

README

• [ ] Similar to intro in the documentation, this needs a lot more big picture so it can stand alone, probably summarize intro page of documentation then link there
• [x] fix small typos, particularly cond install on line 60 since people frequently copy and paste such things.

[codecov-io]

Codecov Report

Merging #26 into master will not change coverage. The diff coverage is n/a.

[bannanc]

Comments from Vickie in PR #41

Intro

• [ ] monte carlo should be capitalized
• [ ] missing period at end of sentence "SMIRKY was tested against"
• [ ] Confusing sentence: "These tools currently use very naive moves in SMIRKS space chosing atom or bond decorators to add to a pattern at complete random."
• [ ] "take away conclusions" is redundant
• [ ] missing a noun in "Then will automatically generate"

Installing

• [ ] typo of "conda" under the OpenEye environment section

API

• [ ] put "layers","ChemPerGraph","mol1", "mol2","#6AH2X4x0r0+0" in code formatting
• [ ] Is there a way you can diagram how these components (ClusterGraph, ChemPerGraph, SMIRKSsifier) relate to one another?
• [ ] I don't fully understand the paragraph under the ClusterGraph example code.
• [ ] reorder with Clustering, SMIRKSify, ChemPerGraph, ClusterGraph
• [ ] "hierarchical list of SMIRKS patterns" -- How is the hierarchy ordered?

[bannanc]

The things I want to add in this PR are still relevant, but this website was basically designed before I had any of the clustering code really working. Considering PRs #45, #48, #50, and #51 I will be able to make the docs more about making SMIRKS based on your input data.

I will be adding larger tests for chemper where it generates SMIRKS based on reference force fields. I can use these as example notebooks in the documentation.

[bannanc]

Some of the points above have already been addressed in previous pull requests, but I have a few other things that I want to do in this update:

• [ ] make sure to reference preprint
• [ ] remove SMARTY/SMIRKY info in the introduction of the README, keep the README to very general information with links to useful information.
• [ ] Use preprint smirksifying example with minimal clusters with two molecules and use the same code.
• [ ] in the example/README.md I want to link to the html examples in the documentation so they could look at them without downloading.
• [ ] update usage examples to use preprint examples
• [ ] Add "History" section to the documentation that has all the SMARTY/SMIRKY information

[bannanc]

I'm going to close this PR. I'm not going to take the time to update the documents any more than I did in the last one. I covered the big things which are that examples should all work (they are tested) and there is reference to the preprint in the README which is currently the best source for understanding the thought process and justification.