Parameter files

Description

The parameter files contain initial estimates of the parameters to be used during the fitting process. These files are provided to ChemEx using the -p or --parameters option.

chemex fit [...] -p <parameter_file> [...]

Parameter files are divided in multiple sections:

• The parameter values under the section [GLOBAL] apply to all residues.
• Residue-specific parameters are specified under sections named after the parameter name, such as [CS_A]. Multiple parameter files can be provided if necessary.

warning

Due to the multidimensional searching feature of $χ^2$ minimization process, it is essential to set a suitable initial value for each parameter to avoid getting trapped in a local minimum.

info

When no starting value is provided in the parameter files, a default value is assigned as the initial value.

Example

Here is an example parameter file:

parameters.toml

[GLOBAL]
PB     = 0.015
KEX_AB = 70.0
TAUC_A = 10.0

[CS_A]
13N = 108.207
26N = 115.711
28N = 113.882
29N = 115.318
33N = 115.636
37N = 116.159
41N = 114.635
42N = 113.525
43N = 108.876
50N = 107.855
52N = 111.358
55N = 128.301
59N = 116.388
66N = 119.429
67N = 114.454
68N = 120.595

[DW_AB]
13N = 4.0
26N = 5.5
28N = 6.5
29N = 6.0
33N = 4.5
37N = 5.0
41N = 6.0
42N = 6.0
43N = 12.5
50N = 8.0
52N = 8.5
55N = -6.5
59N = 6.5
66N = 4.0
67N = 8.0
68N = 4.5

tip

Setting model-free parameters (e.g. TAUC_A) is a simple way to obtain initial estimates of relaxation parameters (e.g., R1_A, R2_A, etc.). For every 2.6 kDa molecular weight, the overall tumbling time is approximately 1 ns at T = 300 K for biomolecules in H₂O. Assuming similar molecular structure at different conditions, the overall tumbling time is proportional to η/T, where η is solution viscosity and T is temperature in Kelvin.

Setting bounds

You can set upper and lower bounds to any fitting parameters by replacing the initial value by a list of three elements:

parameters.toml

PARAMETER_WITH_NO_BOUNDS = <initial_value>
PARAMETER_WITH_BOUNDS = [<initial_value>, <lower_bound>, <upper_bound>]

Such boundaries can help to prevent parameters wandering off to unrealistic values to minimize $χ^2$. However, one should be careful not to set too stringent boundaries either, as this can result in convergence problems. Certain minimization algorithms (e.g. differential_evolution) require finite bounds on all fitted parameters.