Introduction

  To use differentiated cells in a laboratory in a clinical setting, a protocol is essential that is used to generate a reproducible and sufficiently large number of desired cell types^[1]. Every cell has its appropriate culturing and gene manipulation protocols. Human operators introduce an added layer of variability that is difficult to control for, since each person does cell culture a little differently, and the protocols used to manufacture regenerative medicine products are often complex^[4]. In recent years, advanced imaging techniques and analyses have improved the reproducibility of cell culture protocols used in cell fabrication in regenerative medicine^[5]. This suggests that minimizing human error and variability can significantly improve biology experiment outcomes. One way to do this is to mechanize and automate the cell culture process. Replacing difficult and repetitive manual processes with machines makes an important contribution to reducing variability arising from cell culture protocols. In fact, it has previously been demonstrated that automating methods using robotics is an efficient way to reduce human error and worker-dependent variability in the laboratory^[6]. From this point of view, we have built a SynBioBot to realize the above logic by modularizing and automating complex cell culturing and gene manipulation protocols.

System Construction

  Due to its high degree of freedom, the six-axis robot arm allows free movement like a human in three-dimensional space. Robotic arm already demonstrates its capabilities in service, industrial automation processes, surgical operations, and even in 3D printing^[7].

 Based on the above possibilities and the following advantages, we chose the six-axis robot arm as the robot for our project.

 Because cells are sensitive to environmental stimuli, it can be quite difficult even for experienced researchers to produce the desired cell types in a reproducible, sufficient number of cells through cell culture. This means that the culture conditions must be carefully removed. So we've listed the instruments that we need in human cell culture and the elements that can lead to errors, and we've removed some of the unnecessary equipment and some of the elements that make the protocol less reproducible.

 Based on the essential instruments listed, we divided Robot's Working Space into nine spaces: Through this arrangement, the movements required for synthetic biology experiments can be performed, and efficient movements are possible.

Aspect of Experiment

  We conducted various human vs robot experiments to see if we could get real meaningful data through SynBioBot. First, the following Human vs Robot Data could be obtained after pipette calibration. It shows a similar level of error to humans as follows, and even more precise than humans in 25µL pipette (Fig 5). But in case of 1000µL pipette, human result was better.
(a)

(b)

(c)

 Next, looking at the data from synthetic biology experiments, it can be seen that a similar number of cells were obtained from the seeding results of robots and humans after cell thawing (seeding). Even looking at the Light microscope results, no significant difference was found in both cases (Figure 6.a). From the results of the α5 lentiviral transduction, it can be seen that more EGFP was expressed in the automated process, and the gene was successfully introduced (Figure 6.b).

(a)

(b)

 To sum up, the results of experiments conducted by SynBioBot were successful in terms of biology. However, in terms of operating time, the experiment time of SynBioBot took about three times longer than that of humans. This can be supplemented by increasing the speed of the robot when moving the path. We will improve the project in terms of fast speed and precision and accuracy. You can see more results about Wet lab on our Result page.

Aspect of Precision & Accuracy

 To prove the versatility and usability of SynBioBot, the accuracy of the robot itself is also a very important factor. Therefore, our team quantitatively considered the accuracy and precision of the robot. To investigate the robot's acuity, we analyzed the frequency of fatal errors. In addition, in terms of the robot's precision, we measured how far the specific coordinate value deviates from the ideal value in performing the motions.

1. Aspects of fatal error

 The fatal error of the robot to be dealt with in this part is an error that is difficult to perform further experiments after the corresponding error occurs. This error is a factor that affects the accuracy of SynBioBot. Therefore, we investigated these parts by performing experiments repeatedly. In addition, if the experiment was interrupted due to the occurrence of an error, we recorded in which motion the experiment was interrupted.

1.1 Cell seeding

 First, the number of errors in the cell seeding experiment is as follows.

1.2 Subculturing

  In the subculturing experiment, one error occurred. A more detailed analysis of this is as follows.

  As can be seen from the table, an error occurred once in the cap open/close motion. This is analyzed as an error caused by an error in the coordinate value of the AruCo.marker obtained initially. However, considering that the high accuracy of 96% was maintained, it can be evaluated that the experiment was conducted very stably.

1.3 Lentiviral Transduction

  The accuracy measured in the lentiviral experiment is as follows.

  Of the total 35 experiments, no fatal error occurred. As such, only one error occurred in cell-seeding, subculturing, and lentiviral experiments. Through this, it can be concluded that SynBioBot proceeds well in the experiment without errors.

2. Precision

  In addition to accuracy, we analyzed how delicate the robot performs its motion using 'tolerance'. This time, the entire motion was not analyzed, but motions that could affect the precision of the robot were selected and analyzed.

2.1 Pipette motion

  First, we analyzed the motion of inserting a pipette tip to measure the accuracy of the pipette motion. The motion of inserting the pipette tip was performed 30 times with 1000μL and 200μL, respectively. We then measured the success of the motion.

  According to the table, each had a success rate of 96%. Through this, it can be evaluated that the motion of fitting the pipette tip is quite well designed.

  Next, the precision of suction motion during pipette motion was analyzed. When suctioning a solution from a plate using a pipette, the height at which suction is performed is a very important factor in the experimental results. Therefore, we evaluated the accuracy of the suction motion based on the suction position. The reference height was set to a height of 1 mm on the bottom of the plate, and the tolerance was set to 1 mm.

  The table showed that the experiment was conducted very well with 90% precision.

2.2 Plate open motion

  Plate open motion is also an important factor influencing the performance of the experiment. We focused on the motion of putting the plate on the floor after opening the lid of the plate during the plate open motion. How close the lid's position on the floor is to the desired position is a factor that can affect the results of the experiment. Therefore, we have established the best position to perform the experiment. Thereafter, based on the experimental performance, the tolerance was set to 6mm on the x-axis and 3mm on the y-axis. After that, the success rate was calculated by repeating it 20 times.

  From the table, the plate open motion was performed well.

2.3 Cap open motion

  We also analyzed the precision of the cap open motion. I focused on putting the lid down on the floor during the cap open motion. The reason is that after putting the lid down on the floor, you must pick up the lid and close it again on the musical tube. Therefore, we measured how far away the position where the corresponding motion is from the desired position. Based on the experimental stability, the tolerance was set to 4 mm in the x-direction and 2 mm in the y-direction.

  From the table, the cap open motion was performed well with a precision of 90%.

2.4 Suction motion

  When the gripper is used to hold the suction device, the distance between the outside of the suction handle and the metal part of the gripper affects the performance of the experiment. Therefore, we analyzed the precision by measuring the distance. The tolerance was set to +1 mm based on the experimental stability.

  Through this, it could be seen that the suction motion was elaborately well performed.

Aspect of Education

 Due to COVID-19, many students could not go to school and take many classes. Also, students in developing countries don't have the opportunity to be educated in biological experiments due to the absence of laboratories and laboratory instruments. Therefore, Sogang_Korea tried to provide educational opportunities to many people under the concept of "SynBioBot: 6-axis Robotic Arm for Automated Cell Culture and Biological Experiments". The types of online classes mainly consist of watching experimental videos performed by assistants, online simulation, remote control, and virtual reality. However, these classes have various disadvantages, and the most problematic of them is the lack of leading experiments (Figure 10).

We thought SynBioBot could also work as a next-generation educational tool.

Among the factors that evaluate the effectiveness of any education, interest and achievement level are very important criteria. Therefore, we conducted an expected effect survey on biology experiment education using robot arms for Korean high school students, conducted cell culturing and lenticular experiments using robots for university students in Indonesia, and analyzed them in terms of students' interest and achievement.

First, a survey was conducted on the expected effect of conducting biology experimental education with remote robot arms for students at Hansung Science High School in Korea.

27 classmates of EBPH team members participated in the survey, and the following survey results were obtained.

 According to the survey, the majority of students answered that "it would be better to conduct an experiment with a remote robot arm than when they learned the experimental process with only a textbook" for question 1. However, for question 2, the comparison between actual experimental education and experimental education through robot arms, 15 students answered "not sure", and 3 students answered, "I think it will not work". 3 students "I don't think it's going to be good," and 5 students answered, "I think it's going to be good." And 1 student “certainly good”

 Based on these surveys, cell culture experiments and lentiviral transduction experiments were taught to students in Indonesia to find out if there would be actual educational effects. Five students from the Department of Pharmacy and Chemical Education participated, and they performed protocols step-by-step corresponding to cell seeding and lentiviral transduction in real time through our GUI. They were students with experience in learning basic biological knowledge through textbooks. However, through this program, they raised the animal cell HEK293T. They said that they could understand and accept the protocol better by doing experiments through the robot arm. For example, by actually seeing the cells float after adding trypsin, they deeply understood the role of trypsin: detaching cells from the plate before centrifuging.

 All five students said that it was an interesting experimental class and helped them understand the experimental process.

 The results of the survey above prove that remote experiment class with SynBioBot provides students with more qualitative education than learning only with textbooks. In conclusion, it suggests that using SynBioBot for educational purposes is effective.

Conclusion

  Based on the above results, we have demonstrated that synthetic biology experiments can be implemented through our SynBioBot. This indicates that automation of cell and gene manipulation through robots is possible. In addition, it demonstrates that minimizing human error and variability can improve cell culture and gene manipulation results. In terms of education, it confirms that the remotely controllable SynBioBot provides quality education to students. Therefore, we have provided preliminary proof-of-concept evidence for the application of automation and education described on the Implementation page of this project. In the 4th industrial era when ICT and industry converge, the first step in the harmony of cells and robots through SynBioBot suggests the strong possibility of the robot solving technical problems surrounding core synbio technologies inevitably arising from human limitations.

References

1. Sujith Sebastian (2022) Cell Culture: Implementing robotics and artificial intelligence eLife 11:e80609.
2. Carlos A. Tristan, Pinar Ormanoglu, Jaroslav Slamecka, Claire Malley, Pei-Hsuan Chu, Vukasin M. Jovanovic, Yeliz Gedik, Yogita Jethmalani, Charles Bonney, Elena Barnaeva, John Braisted, Sunil K. Mallanna, Dorjbal Dorjsuren, Michael J. Iannotti, Ty C. Voss, Sam Michael, Anton Simeonov, Ilyas Singeç,Robotic high-throughput biomanufacturing and functional differentiation of human pluripotent stem cells, Stem Cell Reports, Volume 16, Issue 12, 2021, Pages 3076-3092, ISSN 2213-6711, https://doi.org/10.1016/j.stemcr.2021.11.004.
3. Genki N Kanda, Taku Tsuzuki, Motoki Terada, Noriko Sakai, Naohiro Motozawa, Tomohiro Masuda, Mitsuhiro Nishida, Chihaya T Watanabe, Tatsuki Higashi, Shuhei A Horiguchi, Taku Kudo, Motohisa Kamei, Genshiro A Sunagawa, Kenji Matsukuma, Takeshi Sakurada, Yosuke Ozawa, Masayo Takahashi, Koichi Takahashi, Tohru Natsume (2022) Robotic search for optimal cell culture in regenerative medicine eLife 11:e77007, https://doi.org/10.7554/eLife.77007
4. Sebastian S, Hourd P, Chandra A, Williams DJ, Medcalf N. (2019) The management of risk and investment in cell therapy process development: a case study for neurodegenerative disease Regenerative Medicine 14:465–488.
5. Yuta Imai, Kei Yoshida, Megumi Matsumoto, Mai Okada, Kei Kanie, Kazunori Shimizu, Hiroyuki Honda, Ryuji Kato, In-process evaluation of culture errors using morphology-based image analysis, Regenerative Therapy, Volume 9, 2018, Pages 15-23, ISSN 2352-3204
6. Archibald, P.R.T., Chandra, A., Thomas, D. et al. Comparability of automated human induced pluripotent stem cell culture: a pilot study. Bioprocess Biosyst Eng 39, 1847–1858 (2016). https://doi.org/10.1007/s00449-016-1659-9
7. Kaiwei Ma, Tianzheng Zhao, Longfei Yang, Peng Wang, Jing Jin, Huajian Teng, Dan Xia, Liya Zhu, Lan Li, Qing Jiang, Xingsong Wang, Application of robotic-assisted in situ 3D printing in cartilage regeneration with HAMA hydrogel: An in vivo study, Journal of Advanced Research, Volume 23, 2020, Pages 123-132, ISSN 2090-1232, https://doi.org/10.1016/j.jare.2020.01.010.
8. J. Chem. Educ. 2022, 99, 2, 519–520. Publication Date:February 8, 2022. https://doi.org/10.1021/acs.jchemed.1c01197
9. https://chem.libretexts.org/Courses/University_of_Arkansas_Little_Rock/
   Chem_1402%3A_General_Chemistry_1_Lab_(Baillie)/1.08%3A_Experiment_7_-_Precipitation
10. František Schauer et al 2008 Eur. J. Phys. 29 753
11. J. Chem. Educ. 2018, 95, 3, 393–399