Adaptive artificial evolution of droplet protocells in a 3D-printed fluidic chemorobotic platform with configurable environments

Evolution via natural selection is governed by the persistence and propagation of living things in an environment. The environment is important since it enabled life to emerge, and shapes evolution today. Although evolution has been widely studied in a variety of fields from biology to computer science, still little is known about the impact of environmental changes on an artificial chemical evolving system outside of computer simulations. Here we develop a fully automated 3D-printed chemorobotic fluidic system that is able to generate and select droplet protocells in real time while changing the surroundings where they undergo artificial evolution. The system is produced using rapid prototyping and explicitly introduces programmable environments as an experimental variable. Our results show that the environment not only acts as an active selector over the genotypes, but also enhances the capacity for individual genotypes to undergo adaptation in response to environmental pressures.


Supplementary
. Image analysis. A: Frame as received from the camera. B: The mask used that defined the arena. All the black pixels were ignored. This mask was always the same through all the experiments. C: Results from the Mixture of Gaussians background subtraction. D: Final result with the detected droplets. All the droplets from ``Bottom left'' which were detected there but that don't appear now were disregarded because they were not big enough, or because they were not moving.

Supplementary Figure 5. Arduino Due Shield.
A shield for the microcontroller board based on "Arduino Due" was developed as part of this project with the objective of maximizing the number of "Pololu A4988" drivers used. Up to 22 drivers could be used, which powered up to 11 pumps (each pump used 2 NEMA motors, one for the plunger and one for the valve). In our case, 7 pumps were used (14 drivers). S6 Supplementary Figure 6. Platform set-up. The device is connected to seven pumps, and each of these pumps is connected to one of the inputs: four oil inputs, aqueous phase, acetone, or waste output. First GA run executed to validate the device, with the GA configuration as defined. There can be seen a drop in values between generation 8 and generation 9. This drop can be probably be attributed to the stochasticity of the algorithm, especially considering the low population size (20) and the high mutation rate used (10%). Figure 9. Second evolutionary trajectory of a Genetic Algorithm run using the empty environment. Second GA run executed. The GA configuration is the same as before. In this case the growth flattened out after generation 4, although the final values almost doubled the initial values. Third GA run executed. The GA configuration is the same as before.

Supplementary
Supplementary Figure 11. Fourth evolutionary trajectory of a Genetic Algorithm run using the empty environment. Fourth GA run executed. The GA configuration is the same as before.  Figure 6 in the main manuscript, this linear plot represents the change in the evolutionary trajectories caused by changing the environment between generations. As before, the change was performed during generations and 10 and 11, and as it can be seen, the fitness value dropped by more than half. All the other GA conditions were exactly the same as in previous experiments.  After 10 generations using the empty environment as seen in the picture on the bottom left, the device was swapped by the device seen in the picture on the bottom right. The evolutionary trajectories hardly saw a difference. Also, the device used during the 10 first generations used "natural" PP, while the one in the right used "white" PP. This, this shows that the different types of PP did not make a difference.  Figure 20. In this case, the first 10 generations used a device printed with "white" PP, and the eleventh generation used a device printed with "natural" PP. There were no major differences appreciated, which corroborates our hypothesis that the evolutionary changes are based on the different arenas used, and not in the individual characteristics of the used device. Hybrid GA run in order to analyse the effect of removing the pillars from the environment. The removal was performed between generations 10 and 11. As it can be seen in the evolutionary trajectories, the fitness values grew slightly. S18

Supplementary Figure 23. Control test 4.
Hybrid GA run where the first 10 generations used an empty arena, and from generation 11 onwards an arena with a procedurally generated environment was used. It can be seen that the evolutionary trajectories dropped, but not as much as they did when swapping an empty environment by one using a pillars arena. Also, the evolutionary trajectories recovered and grew again faster than before.
Supplementary Figure 24. Control test 5. Hybrid GA run where the first 10 generations used an empty arena, and from generation 11 onwards an arena with a procedurally generated environment was used. It can be seen that the evolutionary trajectories dropped, but not as much as they did when swapping an empty environment by one using a pillars arena. Also, the evolutionary trajectories recovered and grew again faster than before.  Best from Empty 0 -9 -5.5 Best from Pillars +2.6 0 +1.7 Best from Caves +1 -6.6 0

Supplementary Note 1: Platform bill of materials
-Each device was designed using Rhinoceros CAD software, and 3D printed using the 3D printer "Bit from Bytes" using polypropene (PP) as thermopolymer.
-The syringe pumps used were "TriContinent C-Series". Three of them used 5 ml syringes, and four of them used 500 μl syringes. All of them used the same 3-way polyether ether ketone (PEEK) valves.
-The electronics from these pumps were replaced with custom-made PCBs in order to power them using an Arduino board.
-The stepper driver used to power the syringe pumps were Pololu a4988.
-An Arduino Due was used to control the syringe pumps.
-"IDEX Health Science PEEK 1/8" tubing was used to connect acetone, aqueous phase and waste from the containers to the syringe pumps, and from the pumps to the device.

S25
-Flangeless fitting nuts, 1/8" OD Tubing, PEEK, were used to connect these tubes to the syringe pumps and device.
-"IDEX Health Science FEP Ora 1/16 x 0.20" was used to carry organic phases from reagent bottles to syringe pumps and from syringe pumps to the device.
-Flangeless fitting nuts, 1/16" OD Tubing, PEEK, were used to connect these tubes to the syringe pumps and device, with corresponding cone shaped fitting.
-A Microsoft LifeCam was used to record the experiments.