From: mk_thisisit
The field of protein design has been significantly transformed by the advent of artificial intelligence (AI), marking a departure from traditional biological engineering methods. This shift is likened to moving biology “out of the Stone Age” [00:25:00], [08:47:00].
Pioneering Synthetic Proteins with AI
A breakthrough moment in protein design was the creation of Top7, the world’s first protein designed entirely using a computer [00:00:16], [01:44:00], [01:53:00], [05:13:00]. Before Top7, all known proteins originated from natural evolution, a process spanning billions of years [00:00:09], [01:51:00], [01:52:00]. The significance of Top7 lies in demonstrating that humans can now create completely new proteins with precise structures [01:46:00], [01:58:00]. Although Top7 initially had no specific function, its creation opened the door to designing new proteins for precise applications [01:53:00], [01:56:00]. Experimental confirmation that Top7’s structure matched its computer model was an “extremely exciting stage” [05:32:00], [05:40:00].
AI’s Role in Overcoming Complexity
The space of possible proteins is vast; for a protein 100 residues long, there are 20 to the hundredth power of possible combinations [04:08:00], [04:14:00]. The key challenge is to find the best solutions within this immense space, such as proteins that can cure cancer or break down plastic [04:23:00], [04:32:00].
The ability to predict protein structure and design new ones has become significantly easier thanks to AI [00:00:16], [01:42:00], [01:44:00], [01:46:00]. The solution to the mystery of protein folding was found without a full understanding of the process, but rather through methods that work effectively [07:05:00]. This success is a result of work by many scientists since the 1960s, which led to an extensive database of about 200,000 protein structures [07:22:00], [07:32:00]. This database serves as ideal material for deep learning methods, which analyze the data, identify patterns connecting amino acid sequences with structures, and predict new protein structures with high accuracy [07:38:00].
AI and Human Collaboration
While AI provides powerful tools, human intuition and a human approach are crucial for identifying problems that can be solved by designing new proteins, such as addressing climate change or treating diseases [08:06:00], [08:12:00], [08:17:00], [08:33:00]. AI will not indicate these problems on its own [08:38:00].
Distinction from DeepMind
Despite sharing the Nobel Prize with co-founders of DeepMind, there is a clear distinction in their work. DeepMind focuses on predicting the structure of proteins, while Professor David Baker’s team focuses on their design [11:06:00], [11:12:00], [11:17:00]. These are related but distinct areas, and the protein design work sometimes utilizes tools developed by DeepMind, making their activities complementary rather than competitive [11:23:00], [11:33:00].
The Future of Protein Design and AI
The impact of artificial intelligence is evident in various fields [03:24:00]. In the last decade, AI, specifically large language models like GPT chats, have brought about many unexpected breakthroughs [14:31:00], [14:34:00], [14:40:00].
Applications of Protein Engineering
Synthetic proteins are poised to have a huge impact on technology and in the near future [01:55:00], [06:06:00], [06:12:00], [01:51:00], [01:53:00]. Their potential applications include:
- Medicine: Creating precise drugs for diseases with minimal side effects [06:16:00], and designing universal “Pionki” (perhaps “pawns” or “building blocks”) to strengthen resistance to various threats [06:26:00]. New biological machines at the protein level are particularly exciting [00:41:00], [01:56:00], [13:56:00]. Significant progress in treating diseases and developing more effective drugs is expected [13:30:00].
- Ecology: Helping break down plastic, neutralize pollution, and remove greenhouse gases from the atmosphere to combat climate change [06:35:00].
- Technology: Enabling the creation of innovative materials and advanced measurement systems [06:51:00]. This includes creating sensors that detect various molecules directly in silicon chips and cell phones, connecting proteins with electronics [00:33:00], [01:48:00], [01:50:00], [18:50:00], [19:01:00]. An “electronic nose” could detect many more compounds than a human can sense [00:53:00], [19:43:00]. This research could even lead to transferring the sense of smell to the digital world [18:35:00], [19:11:00].
Challenges and Unpredictability
Predicting the exact future of science and technology, especially in biological engineering, is challenging [01:01:00], [12:43:00], [12:50:00]. Science develops in complex ways, and unexpected events often open new possibilities [13:08:00]. While AI is advancing rapidly, the question of whether it can achieve “consciousness” remains a significant mystery, as the nature of human consciousness itself is not fully understood [23:31:00], [23:36:00], [23:41:00]. However, it is predicted that computers will become increasingly adept at performing nearly all tasks that humans can do [23:06:00], [23:25:00], [23:57:00].