Four ways that ChatGPT affects drug discovery-Industry News-Lindmik Pharmaceutical(Suzhou)Co.,Ltd

（来源：网络）

It is impossible not to hear ChatGPT online. This is a controversial chat robot launched by OpenAI in November 2022. Here, Diana Spencer of DDW discussed that ChatGPT and similar technologies may change the way of drug discovery and development in the future.
Chat Generative Pre-Trained Transformer (or ChatGPT) is a conversational chat robot designed to interact like humans. OpenAI, the creator, provides the prototype program for free during the test phase and encourages users to share their feedback.

On the company's website, OpenAI said: "The dialogue format enables ChatGPT to answer follow-up questions, admit mistakes, challenge incorrect premises and reject inappropriate requests."
According to Wikipedia, "it has the ability to write and debug computer programs, create music, TV dramas, fairy tales and student papers; answer test questions (sometimes based on the test, the level is higher than that of ordinary human testers); write poems and lyrics; simulate Linux system; simulate the whole chat room; play crossword games; and simulate ATM."
This versatility, coupled with the tendency to give wrong answers, has led people to worry that it will no longer be possible to rely on real-time online interaction, let alone the accuracy of information provided on the Internet.
So how does software like this affect drug discovery, an industry that has so far applied the possibility of artificial intelligence (AI)?

1Aacademic research

In January 2023, the journal Nature reported that ChatGPT was listed as the special author of academic papers, which indicates that academic researchers have begun to use the robot in their research. Many journal publishers argue that chat robots cannot be listed as authors and are forced to formulate policies on how to use or not use them in research environments. ChatGPT's ability to write convincing papers is also an area worthy of attention, and has led people to worry that the form of school and university evaluation standards will become a thing of the past. In order to combat academic plagiarism, student Edward Tian created an application that can distinguish the chat robot GPTZero from the text written by humans, which can bring some hope for the future of written evaluation. Using this software to write a paper may not actually give students or researchers an advantage over their peers, but many people find that when it is necessary to study scientific knowledge in depth, the answer of the chat robot is not accurate. Lorenzo Bombardelli, a business developer and scientist in Amsterdam, posted on LinkedIn: "I'm curious whether # AI Monster # chatGPT can really replace hundreds of hours of study time and years of work experience. Since some universities have banned it, I'm afraid that anyone will suddenly pass any exam, I try to put forward a difficult scientific question that requires specific knowledge of # genetics." In response ChatGPT provided a well written and convincing error response to the request of "Suggesting Light Induced Recombinase", proving that AI like this has a long way to go before it can match or imitate a well-read researcher.

2 Regulatory affairs

One way that ChatGPT conducts tests in the field of drug discovery is in terms of regulatory affairs. The Pharmavies website shows that it can be a useful source of information about different national regulations and help companies understand the regulations related to their products. It also suggests that companies can use ChatGPT when submitting applications to regulators. It asked ChatGPT various very detailed questions related to drug regulations and shared detailed responses, but warned that any information provided by ChatGPT should be checked before being included in the application. In general, the author seems impressed and concludes: "The results given by the program are similar to those given by human feedback. If chatGPT is a glimmer of the added value that AI may bring to regulatory affairs, it looks promising. I look forward to experiencing the iterative algorithm program in the future.

3 Computational chemistry

In a recent study, Gaurav Sharma and Abhishek Thakurb of the Department of Chemistry of Michigan State University tested the ability of ChatGPT in the process of drug discovery, especially computational chemistry. In response to many of their requests, researchers found ChatGPT accurate and useful. It can calculate the multiplicity of compounds, generate input files for Gaussian software and find the required PDB (protein database) files. It is also useful for document retrieval, checking plagiarism and writing basic code. However, they found deficiencies in answering complex questions and providing FASTA sequences and ADMET attributes. Sharma and Thakurb believe that ChatGPT can play an important role in identifying and verifying new drug targets, designing new drugs, optimizing drug properties, evaluating toxicity and generating drug-related reports and papers. They concluded: "It is worth noting that ChatGPT is only one of the many tools used in drug discovery. It cannot replace experimental validation and clinical trials. However, by providing a cost-effective way to process a large amount of data and generate new knowledge, ChatGPT can help researchers make more informed decisions and accelerate the drug discovery process."

4 Protein language model

Although ChatGPT may have its limitations, researchers have been studying the possibility of the technology behind it - natural language processing. This method needs to train AI programs to analyze and synthesize proteins, just as ChatGPT can teach it to recognize and respond to language requests. Karen Hao explained in the Wall Street Journal: "These models encode the so-called protein syntax - rules that control which amino acid combinations produce specific therapeutic properties - to predict the letter sequence that may become the molecular basis of new drugs. Therefore, the time required for the early stage of drug discovery may be shortened from a few years to a few months." Various mature and start-up companies are now using this method to optimize known molecules. It is hoped that this technology can find improved methods in areas that have been considered as non-productive so far.

Future prospects of ChatGPT in drug discovery:

1. Identify and validate new drug targets: ChatGPT can fine-tune scientific literature on the following data sets to generate the latest research summary of specific diseases or biological targets. This can help researchers quickly identify new potential goals or better understand the current research situation in specific fields. 2. Design new drugs: ChatGPT can be used to generate new chemical structures with similar functional groups in known drugs such as molecules. This can help researchers identify new lead compounds with a higher probability of success in preclinical and clinical studies. 3. Optimize the drug compound library: ChatGPT can be used to predict pharmacokinetics and support virtual screening in early drug discovery. 4. Evaluation of toxicity: ChatGPT can be fine-tuned according to the toxicity data set and used to predict the expected toxicological data of new drugs. 5. ChatGPT can be fine-tuned on the data set of known drug-like molecules and used to generate new chemical structures with similar functional groups. This can help researchers identify new lead compounds with a higher probability of success in preclinical and clinical studies.

Disadvantages:

Although ChatGPT can become a powerful tool for drug discovery, there are also some potential disadvantages in using this technology: 1. Dependence on data quality and availability: ChatGPT is only as good as the data it has. If the data is incomplete, biased or inaccurate, the prediction of the model may not be reliable. 2. Lack of experimental verification: ChatGPT can generate predictions and assumptions, but it cannot conduct experiments or measure the resources of compounds. Therefore, the prediction made by the model needs to be verified by experiments. 3. Limited understanding of potential biology: although ChatGPT can give the answer like artificial, it does not understand the potential significance of simulating biological systems. Therefore, the prediction made by the model may not always reflect the real complexity of the system. 4. Limited interpretability: ChatGPT is described as other machine learning models, and it is not always clear how the model makes specific predictions. 5. Limitations of dealing with uncertainty: ChatGPT is a deterministic model, which cannot explain the uncertainty of data and prediction. 6. Lack of transparency: ChatGPT is a black box model. It is difficult to understand and explain the internal working principles of the model, which makes it difficult for people to believe the prediction conclusion of the model.

In summary

ChatGPT has great potential and is certainly much more advanced than other chat robots, but it is still in its early stage and may take a long time before we can fully rely on AI chat robots. We must admit the shortcomings and shortcomings of the current software. Sam Altman, CEO of OpenAI, tweeted in December: "It is wrong to rely on it to do anything important now. This is a preview of the technology exploration; we still have a lot of work to do in terms of robustness and authenticity." At present, compared with real-world applications, it may be more suitable for interesting experiments, but as a proof of concept, ChatGPT may be an AI step with great influence in the field of drug discovery.

Reference:

https://openai.com/blog/chatgpt/

https://www.nature.com/articles/d41586-023-00107-z

https://www.bbc.co.uk/news/world-us-canada-64252570

https://www.pharmavibes.co.uk/2023/02/03/chatgpt-a-chatbot-with-a-myriad-of-potential-uses-possibly-including-regulatory-affairs/

https://www.researchgate.net/publication/367615640_ChatGPT_in_Drug_Discovery

https://www.wsj.com/articles/how-ai-that-powers-chatbots-and-search-queries-could-discover-new-drugs-11670428795

https://chemrxiv.org/engage/chemrxiv/article-details/63d56c13ae221ab9b240932f

[Copyright Statement] Some of the pictures, text and fonts in this article are from the Internet. The copyright belongs to the original author and is only for learning and reference. It is prohibited to use them for commercial purposes. In case of infringement, please contact to delete them.

LINDMIK PHARMA

SIGN UP

LINDMIK MAP

OUR CONTACT