f(t) = Bt + CThis can be used when the data are linear (approximately).
f(t) = At^2 + Bt + CThis should be used when only the initial data points are to be analysed to estimate an initial rate.
f(t) = alpha[1 - exp(-beta.t)] + CThis can be used when initial rates and final horizontal asymptotes are to be estimated from the data.
f(t) = Vmax.t^n/(Km^n + t^n) + CThis can be used when final horizontal asymptotes (also initial rates if n = 1) are to be estimated from the data.
f(t) = Pt + Q[1 - exp(-R.t)] + CThis is used for initial rates, lag times and/or inclined linear asymptotes as found with lag and/or burst kinetics.
Your data may have a nonzero f(0) baseline, e.g. absorbance or background count may not be zero when t = 0, so a model of the form g(t) + A, where g(0) = 0 and A is estimated, might be appropriate to correct for the uncertainty in f(0).
There could be uncertainty in the point t = 0, e.g. if the quenching time is not known exactly when a model of the form g(t - B) with B estimated might be required.
Models like f(t) = g(t - B) + A to cover both cases cannot be used since they are overdetermined, i.e. phi(A, B) = 0.
When t = 0 the models provided are equal to C, i.e. f(0) = C so you can have a best-fit curve through the origin or else estimate the extra constant term C. Unfortunately this can make the curve-fitting more difficult and, in some circumstances, can lead to biased estimates or even failure to converge to a useful solution.
Test files are provided so that you can practise and see what
is involved in fitting the various models. Observe the effect
of fixing the zero point or letting the program estimate it.
Note that the program will not accept negative data values.
Model 1
Use any linear data set such as line.tf1.
Model 2
Use inrate.tf1 which is an example of gently upward
curving data.
Model 3
Use inrate.tf2 which is typical data for the movement
of solute into or from vesicles, with no time lag but
sufficient time to approach equilibrium. For emptying
subtract readings from the initial one to transform
your data into data that increases from zero.
Model 4
Use inrate.tf3. Note that you can fix n at a chosen
value or let the program estimate n. This will only
work if the best-fit n is close to 1 since estimation
of parameters used as exponents depends critically on
starting estimates. Estimate n first, then fix n.
Model 5
Use inrate.tf4 as an example of a lag phase leading
to an inclined linear asymptote.