This web application focuses on enzyme kinetics models, but also supports two binding kinetics models.
Various plots are available to display your data and curves under different transformations. Automating the plotting of data is a significant challenge, so apologies if there are graphical issues. Sometimes hiding the confidence interval helps.
The direct linear plot is a unique plot for the Michaelis-Menten model that puts parameters Vmax and Km as the axes. It visualises all valid pairs of data points as lines representing possible parameter values for that pair. For this reason, data must be provided to use it.
There is an option to show the “true curve”, which is the curve of the model used to simulate the data currently in use. This appears as a black line.
Enable “Advanced Mode” in options to give more customisation options for plotting.
You are able to generate synthetic data based on any supported model. The points on the x-axis to simulate can be adjusted using “Number of Samples”, “Number of Replicates”, “Data x-axis Range” and “Simulation Space”. By selecting a “Dilution Series” simulation space, the sampled points on the x-axis decrease by a selected factor until reaching the max number of samples or leaving the data range.
The noise is added to the y-axis using a normal distribution with a standard deviation selected by the user. For absolute noise, the standard deviation is constant across the x-axis. However, for relative noise, the standard deviation is scaled by the y value at the given point on the x-axis. Often, a significant portion of the normal distribution will be outside the valid domain of the y-axis (e.g. negative velocity). In this case values are clipped to the edge of the domain resulting in a high likelihood of getting values at a particular value such as 0 or 1. Alternatively, by selecting “Truncated Normal” the normal distribution is truncated at the edge of the domain preventing this issue.
You can upload most formats of tabular data (.csv, .tsv, .txt, .xlsx, .xls, .rds). Each row is a sample. Columns are the values used in the selected model, for example velocity and subtrate concentration for Michaelis-Menten. Download some simulated data to see an example.
The following models expects these columns:
| Model | Column 1 | Column 2 | Column 3 |
|---|---|---|---|
| Michaelis-Menten | V | A | |
| Michaelis-Menten with Substrate Inhibition | V | A | |
| Original Michaelis-Menten | V | A | |
| Alternative Michaelis-Menten | V | A | |
| Competitive Inhibition | V | A | I |
| Uncompetitive Inhibition | V | A | I |
| Non-competitive Inhibition | V | A | I |
| Mixed Inhibition | V | A | I |
| Ternary Complex | V | A | B |
| Hill | V | A | |
| Ping-Pong | V | A | B |
| Binding Kinetics | FB | P | |
| Quadratic Binding Kinetics | FB | P | R |
You can fit any model to data using nonlinear least squares. The two most common issues with this are 1) poor starting parameters and 2) fitting with noiseless data. You can try adjusting the parameters manually before fitting the model algorithmicly. By selecting “Add minor noise” perfect noiseless data can be fit using nonlinear least squares.
The Michaelis-Menten supports alternative fitting methods, none of which use starting parameter values. These methods were informed by Fundamentals of Enzyme Kinetics by Athel Cornish-Bowden, 4th edition, and the code is available on the EnzKinet2 GitHub.
Your results from fitting models will be stored in the “Results” tab. For nonlinear least squares, this includes confidence interval values and various statistical metrics. Your results can be exported as a .csv file.
For any queries, feedback, requests or bug reports please get in touch with Haig (haig.bishop@pg.canterbury.ac.nz).