Kinetic Assembly Rates Optimizer

This is a tool that optimizes rate constants for an nmer protein of size n.

Dependencies: Catalyst, OrdinaryDiffEq, DiffEqCallbacks, Plots, Optimization, OptimizationOptimisers, NNlib, ReverseDiff, Roots

Julia Version: 1.10.0

Simplified script

To use this code run the optimize.jl script with the following arguments:

  Arg1: nmer size (n should be an integer)

  Arg2: maximum number of iterations for optimization (should be an integer)

  Arg3: learning rate (should be a float)

  Arg4: Which automatic differentiator to use ("Reverse","Rev","R","rev","r", or "reverse" for AutoReverseDiff() and any other string for AutoForwardDiff()

  Arg5: Which integrator to use (Please specify the integrator in julia format. For example if you want to use QNDF, please input "QNDF()")

Example

Here is an example of how to run the code with a trimer, 1 iteration, .01 learning rate, AutoForwardDiff, and QNDF:

julia +1.10.0 ./optimize.jl 3 1 .01 f "QNDF()"

This code above automatically sets the initial monomer conentrations to 100 and all forward rates to 10.

Using your own Julia script for maximum customizability

ReactionNetwork.jl contains get_fc_rn(n::Int) which will automatically create a rate growth model network of size n based on input. You can use the code like so:

trimer = get_fc_rn(3)

optim.jl contains optim(rn::ReactionSystem,tspan::Tuple{Float64,Float64},p_init::Vector{Float64},monomer_conc::Vector{Float64},lr::Float64,iters::Int,AD::AbstractADType,integrator::Any;verbose::Bool=false).

• rn: Your reaction network generated above
• tspan: The time span to integrate over to determine yield for calculating loss each iteration. I recommend going until the start of the kinetic trapping or just before kinetic trapping occurs. The longer the time span the longer it will take.
• p_init: Your initial forward rates. Backward rates are calculated based on forward rates. The vector should be of length n-1 where the value at each index i corresponds to the forward rates of species of size i.
• monomer_conc: The initial starting concentrations of each monomer species. The vector should be of length n where the value at each index i corresponds to species Xi.
• lr: The learning rate for Adam optimizer.
• iters: The maximum number of iteration to use for optimization.
• AD: Method of automatic differentiation. Currently only works for AutoForwardDiff() and AutoReverseDiff().
• integrator: Input the integrator you wish to use.
• verbose: Whether to print the initial forward and backwards rates.

Example

Here is an example of running a trimer.

new_params = optim(trimer,(0.,.1),[10.,10.],[100.,100.,100.],.01,1000,Optimization.AutoForwardDiff(),KenCarp4())