Team:UPF Barcelona/Safety
Project
Safety and security
- Always wear face mask.
- Promote frequent the hand cleaning.
- Respect the safety distance of 1.5 m.
- In the common spaces, circulate following the sign recommendations.
- Proper ventilation of the closed rooms.
- Notify any symptom related to COVID-19 and confinement in case of close contact. (This caused several leaves during the project development time).
- Team members vaccination.
- Keep reagents away from eyes and mouth.
- Use safety items as gloves and lab coat.
- Use eye protection when looking at blue light.
ARIA's lab
Our wet lab team worked for 2 months in the Translational Synthetic Biology Lab, which belongs to the Universitat Pompeu Fabra and is located at the PRBB building in Barcelona. Before starting lab work, our instructors and PIs prepared the team by giving us some lectures about lab safety, general rules and potentially dangerous situations.
The lab where we worked is Level 1 biosafety. We had a lab bench for the team (see Figure 1), and shared other spaces like benches with Bunsen burners for sterile conditions, air flow cabins, incubators, plate readers and much more with the other lab members. So we team worked with them in order to follow safe and respectful procedures when going from one place to another and sharing materials.
For more images of wet lab spaces and wet lab team working see our gallery page. And for further detailed information about lab security specific protocols, downloadable documentation is available in: evacuation, lab safety rules and transfer of biological samples.
Reagents and strains
The E. Coli strains used for our experiments are Competent Cells NZY5α and BL21, which are not pathogenic. For this reason, the main security measures were wearing gloves, lab coat and lab glasses when doing experiments with them in order to avoid possible skin or eye contact, as well as to avoid culture contamination.
One important aspect to highlight is that gloves need to be taken out when working near the Bunsen burner, since it is much easier to get burned by wearing plastic gloves.
Regarding the reagents, we tried to avoid toxic ones since we were inexperienced in lab work. For instance, we used SYBR Safe instead of Ethidium bromide, which has exactly the same function for PCR gels but avoids carcinogenic risks.
However, for one of our protocols (In vitro RNA isolation), some toxic reagents were cautiously used: QIAzol, Chloroform and Isopropanol. When preparing the experiment protocol, the wet lab team did some research about the safe use of the toxic reagents and its possible reverse effects. For example, all the experiment was done in the gas cabinet to avoid inhalation (Figure 2). A detailed explanation for this protocol is available in the In vitro RNA isolation protocol.
COVID-19 protocols
In the actual global pandemic context, in which Spain is still one of the places where COVID-19 makes the government impose restrictions in order to protect the society, ARIA iGEM team joined the effort in order to protect themselves and the others.
We split into work groups maintaining always the same close contacts, this way reducing infection risks. Only one weekly follow up was done all together, being very cautious with all the safety COVID-19 rules and protocols stipulated by both the Spanish government and the PRBB:
For more specific information, read the official COVID-19 PRBB protocol.
Artificial intelligence
A very relevant portion of the developed work from ARIA was based on the computational research, analysis and final assessment of the obtained results, making the most of the promising Artificial Intelligence field.
However, there were some aspects that needed to be taken into account when it came to AI, the one that poses potential risks, ethical questions and limitations.
To sum up with, AI poses a dilemma, since it offers a huge advantage when it comes to data management and boosting of its capabilities together with the Computer Vision field, but it can also make the system have functioning biases as well as low interpretability of some diagnosis that may be considered as ultimate without deeply knowing the roots, origins and implications of its assumption as ground truth. Regarding interpretability, AI models can be categorized into white-box models, such as decision trees, and black-box models, such as neural networks, which is the case we are considering. Compared with white-box models, black-box models have excellent performance, with almost no interpretability.
More about AI safety can be accessed by consulting “Controlling Safety of Artificial Intelligence-Based Systems in Healthcare” [1].
Cybersecurity and user privacy
Besides, there are two main more concerns about our computational approach: the cybersecurity, that can be found in our Business Plan associated slide and in a detailed cybersecurity document provided by Universitat Pompeu Fabra; and the user privacy when making use of the App. In order to solve the last of them, all the associated data coming from a specific patient, would be processed with a Computer Vision system that, once captured, would convert it all into the means of a vectorized result to be sent to a second AI layer, having no troubles of privacy nor data storage at all.
Final considerations
To end up with, in the long run, still some skills from humans are better and more reliable when compared to AI systems: emotions, feelings, empathy, creativity, intuitiveness, instincts or irrationality are far more golden options that a computer-based solution. However, AI accelerates the knowledge, makes the research exhaustive and poses comfort, industrial benefit and more optimized ways to accomplish tasks that imply large amounts of data. Therefore, the proper combination from conscious human tasks together with the promising AI functionalities, may make the most of our project, always taking into account that every single nuancé should be placed into the safety margins.
User protection
Other important security aspects needed to consider are the ones concerning our final product. On the one hand and in order to get to produce our final kit, clinical assays would be needed, several of them with biological samples provided by a medical center. In those cases, it would be crucial to respect the data treatment and protection policies following the center guidance and to respect the ethical committee requirements as well as to ensure confidentiality for clinical study participants. An example of a clinical trial guide is available in this document, established by the Spanish Agency of Medicines and Medical Devices.
On the other hand, the kit that would be finally created by ARIA should include a client information leaflet and avoid risky reagents, being 100% safe for the clinician. The leaflet would include safety instructions for the product’s use, pointing out not only all the possible dangers taken when handling it but also how to keep away from them. Some important example aspects that we know in advance will be included in those instructions are:
A more extensive and detailed manual will be prepared when ARIA’s kit development is completed. It is considered in our business plan.
References
[1] Davahli, M. R., Karwowski, W., Fiok, K., Wan, T., & Parsaei, H. R. (2021). Controlling Safety of Artificial Intelligence-Based Systems in Healthcare. Symmetry, 13(1), 102. doi:10.3390/sym13010102
