General improvements using eVOLVER as a turbidostat

Dear all,

Since we have acquired our eVOLVERs, we have been working a fair bit on adapting them to our needs and changing a number of procedures that seemed to help the reproducibility and robustness of our experiments. Below is a list of things that we have changed, comments etc…We are working on putting together something very detailed, but as always, time is limiting and we thought we should share the principles already so that we can help you out if you face some of the challenges that we encountered. So if you have any particular question regarding any of these notes, please contact me. Note that all this work is from a PhD student in my lab, Daniel Garcia Ruano, who is our eVOLVER master.

First, all these changes were made to optimize our experimental conditions. We use fission yeast cells, growing at 32 C, and keep them in exponential growth by setting the lower and upper OD in eVOLVER as 0.3 and 0.4, respectively (given these values, we exclusively use OD135). Importantly, we also perform experiments over several weeks. So not all this may be relevant to short-term experiments.


We did work a fair bit on the calibration method (OD) in order to improve the stability and, importantly for our experiments, to have the eVOLVER values as close as possible to what our spectrophotometer indicates.

First, we noticed that in our OD range, the glass vial has a strong impact on the measurements. Even if the normal calibration method (using a rotation of the tubes) may average variations in vials over the 16 tubes, we found this to still introduce significant errors in our later measurements. So what we are doing now is put a vial at each position at the beginning of the calibration and then add cells manually in each vials to the desired OD values that define the calibration range. This way, we still go through 16 measurements, but we never move the vials. The only thing is that we must then edit the calibration file to permute the measurements, as eVOLVER believes that we are rotating the vials (we have a script for this). As a result, there is no change in the input from the vials between the points and our calibration curves are much better. With this protocol, any difference in signal is solely the result of changes in OD of the culture. This improved our data a lot.

Then, to be as true to real OD as possible during the actual experiments, we do not perform a calibration and do many experiments with it. We recalibrate the machine prior to each experiment, with the cells that will be used in each vial for the experiment (we just flush the cells used for the calibration using the pumps, and any left over is OK as these are the same cells that are used for the assays). This way, we do our experiments with the same glass vials that we use to calibrate. This is a bit time consuming but has strongly improved our experiments and is really worth it when performing long-term assays.

Of note, given the strong impact of the glass vial on the results, we are trying to move them as little as possible during the calibration and subsequent experiment. For this, we have designed new caps that lock the vial in a given position in the smart sleeve (see below).


For the experiments, we have changed how the OD is calculated. We are now working only with raw data, until the last calculation. Basically, this allows us to compare our raw data with our raw 0, calculate the difference (which is the result of the cells), and report that difference to the calibration curve, starting form the rawcal0. This way, we get the OD quite precisely and, importantly, really close to what our spectrophotometer tells us. This is essential to compare the generation times of different cultures in different smart sleeves.

Other mods

  1. Software:
  • There was a glitch in the software when performing very long experiments, as the loading a large datasets was slow, giving rise to abnormal dilutions. This has been solved.
  • The software now tracks media consumption based on pump operation, allowing you to have an idea of what should be left in your bottles. We have made an interface that allows you to indicate the volume of the bottle you are plugging and the vials to which it is connected. It then calculates consumption and what’s left in the bottle.
  • We have added a growth rate plot to the graphing tool.
  • We have made a script to delete the old calibrations that are not required anymore.
  • We have made a script to reset the blank after changing a vial mid-experiment (in case of a contamination, for instance)…allows to limit the impact of the glass vial when a new vial is introduced in the middle of the experiment, which has not been used for the calibration.
  • We have made a script to get the OD at a given time point (rather than going in the OD file), allowing us to easily compare during an experiment whether there are differences between ODeVOLVER and OD spectrophotometer.
  1. Hardware:
  • We have made a new cap system that has 2 advantages: it prevents the vials from moving, and it protects the machine, as any leak or overflow cannot go between the vial and the smart sleeve (CAD file can be sent).
  • We have built a complete casing in PMMA that protects the entire machine from leaks (AI file can be sent), including rubber pads that make sure that no medium can damage the machine.
  • We are using laser cut silicone joints to seal the cap with the vials.
  • We are also now using individual scales for each bottle of medium. These scales are connected to a system that sends us messages if the volume (weight) of remaining medium in each bottle is below a user-defined value. This goes through your mobile phone…this was developed together with a local company, so I am not sure how easy it would be to adapt it outside of France.
  • We are now using the new resistor pack from Fynchbio, as it does significantly improve the dynamic range for the OD 0.3-0.4 window that we use. This requires a vial-per-vial optimisation of the IR LED power. To this end, we performed several calibrations with different powers (changing by 1 or 2 units up or down) and determined the optimal settings for each smart sleeve.

That’s about it ! All this will be put together and made available here (some scripts are already there in fact). But if you need anything that is not “officially” available, please contact me, as this is just a matter of time for us to compile everything.