There are many iGEM teams in Nanjing, China. To build up a platform where communication and collaboration of local teams could be facilitated, NJU-China 2021 organized Nanjing iGEM Association (NIA). On the 27th of May, the 2023 NIA Meetup was held in Nanjing University, hosted by NJU-China and attended by other eight teams from local universities and high schools. We also invited Bao Yuhan from iGEM HP committee and Xu Yilong, 2023 iGEM Ambassador for Asia, as guests to present lectures to all members present and give suggestions to the reports.
1. Lecture
At the beginning of the event, Mr. Bao delivered a speech themed "How to evaluate the progress and quality of HP" through online meeting. He clarified the principles and concepts of HP, emphasized the close integration of HP activities and project which featured a two-way influence, and reminded everyone to know that the motivation, reason and harvest should be sufficient.
2. Team Report
In team report section, each team introduced their projects. We shared our project design about AI modified yeast promoter. Specifically, we select promoter regions of genes whose strength and weakness rules are still not fully deciphered, and train AI models to learn their underlying mechanisms and generate promoter sequence with required expression levels. In the future, we hope to conduct specific experiments to apply the AI model to real-world challenges. After the introduction, Dr. Xu Yilong raised questions and our speaker carefully explained them.
3. Communications
During the exchange process of the meeting, teams were free to communicate with the other group members they were interested in. Some discussed design considerations, some discussed how to organize human practice activities, and some shared their own experiences, reminding other teams to avoid missteps. We proposed possible problems with respective projects and discussed solutions together. Especially, we had a deep exchange on modeling with CPU-CHINA and got helpful suggestions from them. We also talked with NNU-CHINA and NJMU-China about how to organize integrated HP and education activities, and got a lot of experience from NJMU-China.
The Conference of China iGEMer Community (CCiC) serves as an important platform for communication and collaboration among iGEM teams in China. To strengthen the connection with other iGEM teams from brotherly universities, and to showcase our team's current achievements, we participated in the offline 10th CCiC held at Hainan University from July 8th to July 11th.
1. Lecture
The conference featured renowned experts and scholars in the field of synthetic biology, including Ms. Zhang Nan, Vice Chair of iGEM Global Development, who provided insights into the history and evolution of iGEM. Professor Wang Baojun from Zhejiang University presented on synthetic gene circuit design and innovative applications. Additionally, Professor Chen Xing from Peking University introduced the topic of glycosylation from a chemical perspective. We got a deep insight into the leading edge of synthetic biology research from the lectures.
2. Team Report and Communications
During the presentation section, we provided a brief overview of our project's design and progress. After the report, we engaged in in-depth discussions about our model with SJTU-software, CPU-CHINA and Tongji-Software, and received lots of detailed suggestions about the training process. They recommended we focus on DREAM challenge, a competition about predicting gene expression using millions of random promoter sequences, which might provide useful information we need. They also reminded us to consider measures of fluorescent protein expression, such as whether the ratio to cell density was taken into account. It inspired us to choose a dual fluorescence reporting system to characterize protein expression to rule out extraneous factors.
Since we focus on the application of AI in synthetic biology, we are eager to explore this topic with other teams to gain new perspectives and to further our understanding about it through discussion. With this purpose, we participated in the online seminar hosted by SYSU-Software on August 9, which themed Synbio Plus AI: Emerging Complexities (in Interdisciplinary Views). We had a discussion centered on the crossing field of Artificial Intelligence and synthetic biology with other five teams, including SYSU-Software, Tsinghua-A, Tongjing-Software, DUT-China, and JLU-China. During the communication, we received some inspiration and validation of our existing ideas in conducting experiments. AI for biology has not only attracted the attention of scientists who have been engaged in scientific research for many years, but also gained a stronger interest from undergraduate students, who are more eager to get started and practice. We found that most teams had generally used the traditional paradigm of active learning, which embodies the high threshold of AI and the difficulty of getting started, specifically reflected in the challenge of obtaining large sample data, building AI models from scratch, and the high error rate of small samples. This also reflects the significance and promotion value of our transfer learning paradigm. After our discussion, everyone generally tended to agree that AI is a practical, revolutionary, and disruptive tool for biologists, who lack the energy and necessity to build large models or optimize algorithms from scratch.
Furthermore, the communication between the teams raised our attention to the issue of open-source data and algorithms. Everyone acknowledges themselves as beneficiaries of open source, and supports to apply open-source principles to data and algorithms. We also agreed and understanded that in order to obtain benefits, it is possible for some companies to keep their data or algorithms secret, or only partially disclose them. However, for scientific research, especially for iGEM, data and algorithms must be fully disclosed, because only in this way can the continuous progress of this discipline be promoted for a long time. In addition, it also makes a difference to facilitate the migration and transformation of other biologists.energy and necessity to build large models or optimize algorithms from scratch.
In August and September, NJU-China collaborated with other ten iGEM teams all over the nation and from India to carry out science popularization activities for exceptional children, including hearing-impaired children and kids with autism, who are often not suited to learn in regular schools. We hope to provide them with more diverse ways to learn new knowledge and have an insight into a wider range of fields.
1. Nanjing-China
Since we are from the same university, NJU-China have multifaceted communication with Nanjing-China throughout the project, from which we received some new inspiration as well as validation of our existing ideas.
We discussed the new medal criteria for this year together and exchanged views on how to organize and conduct human practice events. They shared their human practice planning and advised us to combine it with the experiment circles closely. We reached a consensus that stakeholders from different fields including the public, academia, industry and so on should all be taken into consideration, since they have respective roles to play in different stages throughout our whole project. It inspired us to systematically manage our related stakeholders and figure out the order to communicate with them.
In the later stage of the project, we also communicate with them about how to present our human practice work in Wiki pages. They advised us to organize them in line with the process of project design, instead of sorting by the type of activities or stakeholders. On the other hand, we believed that their projects have high practical application value, and it’s rewarding to consider the benefits in industrial production, obtain the evaluation of their projects from enterprises and try to seek commercial cooperation. What’s more, because we don't have members proficient in front-end development, they provided us with a lot of help and guidance on wiki coding.
In terms of experiments, we shared our high-fluorescent yeast strains which we optimized with Nanjing-China. They co-cultured them with Acetobacter xylinus in their experiment to detect the coating mode. Since Acetobacter xylinus produces a very thick cellulose membrane, it helps to verify the fluorescence intensity of yeast using our AI model to optimize promoters.
2. CPU-CHINA
In the NIA meetup, several iGEM teams from Nanjing attended, as did Teams CPU-CHINA and NJU-China. We presented our project design at the meeting, and CPU-CHINA showed great interest in it because dry lab also plays a large part in their projects. During the free communication after the meeting, we had a heated discussion with CPU-CHINA about model design and built a bridge between each other.
During the follow-up experiment, we kept in touch with them through online meeting to exchange the latest progress. We asked for their views on whether the entire AI can be regarded as model and got a positive response, which is an important reference for us to set the direction of model design. Since the dry experiments of our two teams are both iterative processes, we have encountered the same problem of how to better display them in a way that is both logical and closely combined with the wet experimental design. After many exchanges of ideas, we came to a conclusion that it’s more appropriate to integrate it into each step of the engineering circle after summarizing, and add a link to jump to the model part for detailed information. CPU-CHINA also offered detailed suggestions on engineering and model content of our wiki page.
3. Tongji-software
Tongji-Software develops a gRNA design tool applied to CRISPR virus detection to recommend gRNAs that can efficiently detect the target virus. They use a deep learning framework based on convolutional neural networks to analytically compute the relationship between gRNA sequence, target sequence and gRNA activity. Since the training of deep learning models plays an important role in the project of both our two teams, we had a further communication with each other in 10th CCiC and hope to have a continuous cooperation in the future.
We held online meetings in middle and later stages to share our progress with them. At that time, our dry experiment project's Pearson correlation coefficient could not achieve the expected effect. After communicating with them, it was determined that only fine-tuning would be difficult to significantly improve the model's prediction performance. We could only start from the model itself or the data, which also inspired us to optimize the model structure or replace the pre trained model and research data, indicating the direction for our next dry experiment development.
As Tongji-Software has been focusing on dry experiments and modeling in projects of previous years, we refer to the content and the pattern of organization of their Model page and Engineering page. We acknowledge them for their incredible thoughts and passion for partnership, which make all of our teams benefit a lot.
