ToDo List for issued problems

[shaoqx]

https://github.com/ZJYgrp/enzyme_workflow/issues/11#issuecomment-789395319 https://github.com/ZJYgrp/enzyme_workflow/issues/10#issuecomment-782613999

[shaoqx]

About mutation from small to large: Problem: This may generate some initial structure that contains some inevitable bad contact. (even with rotamer lib) Potential Solution:

• Classification first
• (S->L) Template-based modeling: [ ] create an interface with Swissmodel. [ ] create an interface with AlphaFold2. This method considers the global structure change caused by a S->L mutation

[shaoqx]

General TODO:

• ONIOM pseudo-H label
• Use memory for QMMM template

Next Phase

• Structure Prediction
  • AlphaFill (https://alphafill.eu/)
• Loop refinement (Rosetta? trRosetta?)
• Engine
  • Better Gaussian runner & Batch submission (parallel energy engine job)
  • Interface with engines of QMMM, FE, etc. (ChemShell, Terachem, Q6, BFEE)
• Protonation
  • better implementation for ligand (simple chemical sense, H-bond network) (Maybe get rid of the PDB2PQR, add H by EnzyHTP itself with the information from PROPKA3 which can also give info about substrate H-bonding network.)
  • more strategy around metal
  • try testing
    • AF2 (not working well)
    • H++: Can consider metal's influence in the pKa calculation. (e.g. 1NVG) Can include ligand gives weird protonation and sometimes ignores the user input total charge.
    • molprobity
• Mutation
  • coupled with the protonation problem The target is to optimize in the space of both sidechain conformation and the protonation state. But in practice, we can only do one at a time. The initial structure of mutation depends on the initial protonation state and thus the optimization in the geometry could be limited by such protonation and can not reach the real minima and without a good position of the heavy atom, the protonation state can never be correctly determined. A solution would be mimic the strategy in Rosetta, send some decoy to MC and sample initial protonation state and do optimization.
• MD input
  • MCPB with metal (PDB2FF)
  • artificial residue or bonded ligand
• Docking
  • key water reservation
• Better interface with database?
• Amber Interface
  • pytleap?

syntax details

• unify and support enhanced customization of the tleap input file. (converge to PDB2FF, support edit line)
• Optimize the data structure of class structure. (In the structure level, the iteration problem. Generalize the concept of chains but still maintain the handy separate naming.)
• try using black to format the code.
• try using pickle db as the output format instead of text.
• fix equi cpu requirement. Clue: 1. change solvatebox to oct 2. reasonable restraint (2 step equi maybe?)
• fix antechamber print problem (use run()?)
• support output job log to a file but not stdout
• support continue run when error occurs due to structural problem.
• support put all traj, gaussian out&chk to a scratch path
• merge layer define tool for QMMM and QM cluster
• decoding amber style atom mask (in helper and Class_structure) (Or just use getitem and getattr like method in Class_structure?)
• logging & Config.debug
• replace os.system with subprocess.run (os.popen used to lead to some bugs) https://docs.python.org/zh-cn/3/library/subprocess.html?highlight=subprocess
• support separate MD step functions and a custom MD function.

[shaoqx]

Updates proposed in EnzyHTP 2.0:

distribution

• [x] conda and pip support
• [x] concise quick start up tutorial

structure prediction

• [ ] AF2

preparation

protonation

• [ ] better ways to protonation ligands (test & apply)
• [ ] refine protonation around metal (for HIS or ARG specifically)

mutation

• [ ] Modeller
• [x] PyMol

sampling

MD

• [ ] better support for ncaa
• [x] better support for constrain
• [ ] support MCPB for metal (e.g. P450)
• [ ] interface with CHARMM for lambda dynamic (efficient ddG calculation)

Analysis

• [x] interface with Rosetta
• [x] interface with cpptraj

Test

• [ ] FAcD for 3F substrate, featuring "lego"-like workflow design to solve a real scientific problem. `