1. Autoclave vials with lids on.
   1. For “open” vials, it is not necessary to autoclave the needles.
2. Clean lines with sterile water (20s), and 70% ethanol (20s).
3. Weigh empty vials, note OD of (optional: washed) dense cultures.
4. Fill each vial with 25mL of media. Connect and run efflux lines with cleaned efflux needles. Weigh final culture volumes.
5. Once vials are connected to the fluid lines set up the experiment.

Setting up the experiment:

1. Open up an iterm2 terminal (Command+space iterm2).
2. We use bash instead of the default terminal. Type in bash and hit return.
   1. If this was done right, the prompt should change, and should be preceded by something like (py37) which indicates that the virtual environment has been loaded.
3. In the file browser (“finder”), navigate to Documents/dpu/experiments/.
4. Duplicate the evolver/ folder, and name the duplicate with a descriptive name. Typically YYYY_MM_DD_keywords_separted_by_underscore.
5. Switch back to the terminal window and type in cd ~/Documents/dpu/experiment/NAME_OF_YOUR_NEW_DIRECTORY
6. Start the visual interface to populate the configuration file by typing in streamlit run visualize.py. This will return a localhost IP address with a port. Copy and paste this in a browser window.
7. Populate all the fields, and press ‘Write to configuration file’. NOTE: If you are running a calibration, or if you are not planning on using a calibration file, LEAVE THE “Calibration name” field BLANK!!!
8. Alternative approach: The folder has an example_experiment_parameters.yaml file. Edit the contents of this file and save it to a file called experiment_parameters.yaml in the same directory.
9. If you are running a calibration:
   1. Include the word “-calibration” in the experiment name.
   2. Currently, if the per vial parameter calib_end_od is left at its default values (0), it is assigned a value calib_start_od/10. Please explicitly specify this parameter taking into consideration the sensor readings saturate around 0.075.
   3. Open a new terminal window, load bash, navigate to the experiment folder, and start the calibration by typing in python eVOLVER.py. This will display the calibration settings. When prompted about blanking vials, type n.
   4. The calibration will stop at the last step. Collect the efflux from each tube. Select all the vials from the Rpi touch interface on the eVOLVER and run the efflux for 10s. Measure the ODs. Multiply the ODs with a factor of 1.2 if measuring efflux culture.
   5. Go back to the web interface, update the calib_end_od with the new ODs, write this to file again. NB: The old file will be automatically renamed and time stamped.
   6. Open a new terminal, load bash, navigate to the experiment directory as in STEP 5.
   7. Type in python write_calibration_to_file.py. This will create a CSV file with the name of the calibration experiment, the same as the experiment name.
   8. Next, switch back to the browser and type in the name of this calibration. The real experiment will need to have a different name.
10. When the experiment config file is filled, switch to the terminal and in a new tab (load bash and navigate to the experiment direcotory), type in python eVOLVER.py
11. Sanity check the experiment settings, press n when prompted for blanking the vials, and the experiment should start.
12. When the experiment is running, a new drop down option should be added corresponding to the experiment name, displaying all the plots.

1. The major caveat in passing message to the Rpi with the current design is that the strings are all expected to be 16 values separated by commas. This is currently handled by looping over vials which has a list of 16 integers, and populating lists like volumes, rates, and periods by looking up indices from other data sources like user defined variables. This is pretty low level right now, and there is no real API to arbitrarily control sleeve behavior without resorting to hacks like creating custom log files for history dependent control of sleeves
2. There are low level (arduino-level) event logging requirements which are going to be important to debug. This will impact things like:
   1. Frequency of data logging
   2. Types of data that are logged systematically (pump events in chemostat mode etc)

I am not a fan of this design, where the custom script is loaded in order to first load parameters, but control is then handed back to eVOLVER.py. Custom imports

• Ideally, the config file is read once, and parameters are passed around in some data class or similar. See Issue http:github.com/amoghpj/evolver/issues/1.
• No consistent API in adding custom experiment handling. Need to move away from using per-vial logic in a loop like it is currently setup.
• Define defaults for stirring control that are passed explicitly to each function definition

We can perform an in-vial calibration by diluting out a dense culture by dispensing media of fixed volumes repeatedly. Interestingly, performing a 100X dilution in steps uses far less media than doing it in a single step. The total volume of media consumed as the bolus size → 0 is - volume_vial * log(endOD/startOD).

Below is a visual depiction of the relationship between bolus size, number of steps needed to hit endOD, and the total volume dispensed.