Sensor Overlord

Can your sensor make accurate biochemical measurements?


When you want to make absolute, quantitative measurements with a ratiometric biosensor, we often find it difficult to choose a sensor well-suited for our experimental designs. Or, if we have already choosen a sensor, we usually want to know whether it is accurate enough for our experiments.

For example, it's generally accepted that roGFP1 is a good sensor to measure redox potential in C. elegans cytosol, but that roGFP1-iE might be a better choice for the ER.

This tool serves to quantify that intuition and help you make better-informed choices for your experiments.


This tool is appropriate for any ratiometric sensor with two states. It will be useful if you convert a ratiometric measurement into a real biochemical value, such as the concentration of a ligand or a redox potential.

This tool puts all sensors into three categories:

  • Redox: Sensors can be oxidized or reduced, and their ratio is converted to a voltage
  • pH: Sensors can be protenated or deprotenated, and their ratio is converted to pH
  • Ligand: Sensors can be bound or unbound, and their ratio can be converted to pLigand (-log10 of ligand concentration)
  • Instructions

    To predict whether a sensor is well-suited to your needs, you need to know a few things:

    How precise are your microscopy techniques?

    Your microscope and analysis methods will produce noise in your ratiometric measurements. To determine that noise, run through your microscopy pipeline with GFP or another mock sensor, making measurements that should be consistent with each image (for example, 410/410 with a roGFP, or 410/470 with GFP).

    Then, use those measurements to estimate your relative error. For example, if you find that 95% of your 410/410 measurements are between 1.02 and 0.98, your relative error is +/- 2.0%.

    How accurate do you need your measurements to be?

    For example, you may be trying to distinguish between a pH of 7.1 and 7.3, in which case you will need an accuracy of at least 0.2.

    What are the biochemical and biophysical parameters of your sensor?

    You can generally find estimates of these parameters in the literature. We have also aggregated these estimates in the supplementary notes of the associated manuscript.

    Ready? Get started here:

    Sensor Parameters

    What sensor would you like to use?

    We have loaded a few pre-analysed sensors to get you started. Feel free to choose one of those from the dropdown box.

    Once you're comfortable, select 'Custom', and then scroll to the bottom of the page to input the characteristics of your sensor of choice.

    What is your microscopy precision, and how accurate would you like to be?

    You'll need to determine your microscopy imprecision empirically. For reference, 95% of our observed ratios deviate from their true value by no more than 2.8%, but that can vary from 1% to 4% depending on image analysis methods and experimental conditions.

    Accuracy can be determined as you see fit!

    What wavelengths are you using to make your measurements?

    You may find that adjusting your wavelengths or narrowing your band size may affect the ultimate accuracy of your measurements.

    Suitable Range


    Predicted Inaccuracy Plot


    Custom Sensor Input

    SensorOverlord needs Rmin, Rmax, delta, and midpoint values. It can either determine some of those from an uploaded spectra file ('Upload Spectra' tab), or you can provide those parameters ('Input Characteristics' tab).

    Thanks for visiting SensorOverlord!

    For more information, here are some resources:

    Full SensorOverlord package

    SensorOverlord is a full R package that extends the functionality that you have seen in this application. For instructions on how to use and install the SensorOverlord package, see the associated github package.

    Scientific Manuscript and Citation

    If you use SensorOverlord in your work, please cite our manuscript as follows:

    Stanley, J.A., Johnsen, S.B. & Apfeld, J. The SensorOverlord predicts the accuracy of measurements with ratiometric biosensors. Sci Rep 10, 16843 (2020).

    The paper includes a much more rigorous overview of the SensorOverlord package, as well as its applications and derivations. It is publically-available here via Scentific Reports.


    If you have absolutely any questions related to SensorOverlord, or just want to get in touch, please feel free to reach out to me (Julian) at julianst [at] mit [dot] edu, or to my PI (Dr. Javier Apfeld) at j [dot] apfeld [at] northeastern [dot] edu.

    Best of luck with all of your microscopy endeavors!