# Data Calibration

This page is *completely unrelated* to Smoldyn, except that it was also written by
Steve Andrews.

It performs data calibration using a new method, called the 1-step method. This contrasts the conventional 2-step method, in which the first step is to use measurements of standards to estimate a calibration curve and the second step is to calibrate the data using the calibration curve. The 1-step method uses information from repeated measurements of unknown samples to reduce the effects of measurement error, especially in standard measurements. This method is particularly useful when samples are analyzed in "batches," where a batch might be a protein gel or a series of analyses; by definition, calibration parameters are the same for all measurements in a single batch and different for measurements in different batches.

This method was recently submitted for publication in PLoS ONE, under the title of "Maximum Likelihood Calibration for Samples that are Quantified in Batches", by Steve Andrews and Suzannah Rutherford.

The measurements are assumed to be accurately represented by the model:
*y _{i,j,k}* =

*α*+

_{i}*β*+

_{i}x_{j}*ε*, where

_{i,j,k}*y*is an uncalibrated measurement,

_{i,j,k}*α*and

_{i}*β*are batch-specific calibration parameters for batch

_{i}*i*,

*x*is the amount of analyte in sample

_{j}*j*, and

*ε*is Gaussian-distributed measurement noise with standard deviation

_{i,j,k}*σ*for that particular measurement.

You can try this method here using our data, random data, or your own data. You can also download the Python source code from here. If you encounter problems, e-mail Steve Andrews.