Does an exponential model fit my data? I am measuring accumulation of a fluorescent-tagged protein at a particular location within a cell over time. In previous experiments that I have performed, I see a standard exponential distribution where the fluorescence intensity reaches a plateau, however in my current experiment, I see a distribution as shown below:

What is the best model to use for this data? Should I use two separate exponential models, one for the increase in intensity up to the peak and one for the decay phase, or is there another statistical model for this type of distribution. Thanks!
 A: I echo @glen_b 's comment. Don't just fit random functions, think about it. What causes the signal to rise and fall? Examples:
rate of transcription
rate of translation
rate of transport
percent transported to your region
rate of degradation
Is quenching of the signal occurring due to high concentration, incorporation into a complex?
Is the protein denaturing for some reason (lysozomes?) thus affecting the signal?
Are there a plausible negative/positive feedbacks occurring?
What are you normalizing to? Are you sure this isn't rising after ~ 2minutes?
What is new about this time? This seems to be an important hint.
Also many curves people fit are artifacts of averaging, there is a long history of people complaining of this in the learning literature. The correct thing to do is fit a curve to each time-series that has been derived from assumptions that people in your field will find acceptable. There is currently not enough info in your question to provide a reasonable answer. You have a nice smooth curve here, don't miss the opportunity to solve it with a series of differential equations.
