Among the utilization of AI (Artificial Intelligence) in drug discovery, there are high expectations for the utilization of AI in the initial stage of drug discovery, i.e., target search, or the hypothetical stage in which target candidates are selected and research, and development begins.
Hypothesis generation is the creation of a disease mechanism that predicts the relationship between a target molecule and a disease, and hypothesis here is one of the most important elements utilized in the process of drug development from the most upstream stage, target discovery, to market launch.
FRONTEO's Drug Discovery AI Factory, which utilizes KIBIT, an in-house developed natural language processing AI engine, is expanding a new service that supports the efficiency, acceleration, and improvement of success probability of drug discovery research by utilizing AI specialized for hypothesis generation, which is required in various phases of drug discovery.
What is Hypothesis Generation?
In general, "hypothesis generation" in scientific research refers to an approach that proposes a new hypothesis or idea for an unsolved problem or unknown phenomenon.
There is also a "hypothesis-testing" approach to research, where hypothesis generation is the process of exploring new ideas and directions, while hypothesis testing is the process of testing a known hypothesis or theory. In scientific research, progress is made through a combination of these two approaches.
|The hypothesis-generating approach is used in scientific research to generate new ideas and theories. Researchers form hypotheses based on existing knowledge and observations, and then conduct experiments and tests to find new knowledge and theories. Hypothesis generation is an important step in a scientist's quest for new directions and theories with the aim of generating new knowledge from known information.
|The hypothesis-testing approach is used to ascertain whether an existing hypothesis or theory is correct. The researcher conducts specific experiments or observations based on an existing hypothesis and then uses the results to test the validity and accuracy of the hypothesis. This approach emphasizes scientific rigor and is useful for establishing the reliability of existing knowledge and theories.
Challenges and Current Status of Drug Discovery
- Hurdles in Generating Hypotheses for New Drug Development
While the success rate of new drug development is lower than before, the cost of research and development is rising acceleratedly. In " Target identification," which is the start of drug discovery, the selection of candidate targets and the estimation of how they will lead to a disease, in other words, the prediction of the disease mechanism, are particularly important.
The search for target molecules is one of the bottlenecks in drug discovery research, as it involves many issues such as researchers' own biases and reliance on personal knowledge. In addition, the prediction of disease mechanisms to support this process requires significant human resource costs, and it is extremely difficult to find an AI drug discovery support company in Japan or anywhere else in the world that can provide hypotheses on disease mechanisms, etc., even if they are requested by outside parties.
"Drug Discovery AI Factory"
Generating Hypotheses as a Starting Point for Drug Discovery
In response to the challenges of drug discovery, FRONTEO launched the Drug Discovery AI Factory, which supports drug discovery with the power of AI. It is a platform that generates and continuously supplies hypotheses to produce new ideas and directions in drug discovery research.
The hypothesis to initiate the development of a new drug includes,
-Target molecule: A gene or molecule etc., that is predicted to be associated with a disease.
-Disease mechanism: How the target molecule is related to the disease, etc.
-Patient information: Genome information, diseases, symptoms, etc., of the target patient group.
-Safety information: Toxicity of the drug candidate, etc.
-Feasibility: Proposed experimental model, etc.
FRONTEO believes that all of these factors need to be considered.
“Drug Discovery AI Factory” makes it possible to generate hypotheses in a single step for everything from candidate target molecules to analysis of mechanisms of action, genome information on related diseases, and proposals for new drug safety as well as its experimental models, by maximizing the utilization of AI, which has the strength of analyzing vast amounts of data.
Comprehensive and Unbiased "Target Molecule Search and Selection" for New Drug Candidates Utilizing AI
To select highly novel target molecules, FRONTEO's AI analyzes literature information from more than 30 million reports in PubMed*. It finds highly novel target molecules that have not been found by existing approaches such as gene expression analysis and GWAS, such as genes predicted to be associated with diseases.
*Biomedical article database operated by the National Center for Biotechnology Information, U.S. National Library of Medicine.
AI Enables "Hypothesis Generation" for New Drug Candidates, Biologists so far as Propose Hypotheses
On the other hand, a list of highly novel target molecules is not enough to advance the drug discovery process. At FRONTEO's Drug Discovery AI Factory, which specializes in hypothesis generation, biologists with expertise in AI and drug discovery decipher hints that lead to hypotheses from multiple AI-based analysis methods and corroborate them with background information. The probability of success is dramatically increased by further narrowing down from the highly novel targets obtained by AI.
Themes on Which Hypothesis-generating AI
Can Generate Hypotheses for New Drug Development
Target Identification and Hypothesis Generation
Disease-related target molecules can be searched for in tens of millions of publications by AI, and highly novel targets can be discovered in a short time. AI can also draw a network of related genes for the entire disease in just 10 minutes, resulting in new discoveries.
Drug Repositioning (DR) and Hypothesis Generation
Drug repositioning (DR: redevelopment of existing drugs) is a method of developing existing drugs or drugs whose development has been discontinued as treatments for other diseases by discovering new efficacy. By utilizing AI to cover all types of paper information, even information that researchers could not fully grasp is analyzed to generate new insights and ideas, such as unknown connections and possibilities.
Suppressor Mutation Genes and Its Hypothesis Generation
We can also search for suppressor mutations for diseases caused by loss-of-function mutations. By drawing a unique gene network that includes AI-predicted connections, we can understand the network between genes and its strength, and predict suppressor mutant genes.
“DD-BKM" Supports Hypothesis Generation
for Drug Discovery with 5 Analysis Methods
Drug Discovery Best Known Methods (DD-BKM) is a fusion of FRONTEO's natural language processing AI "KIBIT" and drug discovery researchers. There are currently five analysis methods, which researchers combine and analyze to acquire clues and generate hypotheses that would not be thought of by humans alone.
-Common-Unique Pathway Analysis ~Comparing disease networks to find common and unique pathways~.
-Two-dimensional Mapping Analysis ~AI analysis of medical papers and mapping by conceptual similarity~
-Vector Addition Analysis ~Finding important factors using a natural language approach~
-Multifaceted Analysis ~Comprehensive analysis of multidimensional items to analyze the potential of a gene as a drug target~
-Virtual Experiments ~Virtual knockout of genes and simulation of pathway changes~
AI-savvy Biologists and in-house Developed AI
In order to analyze information and derive innovative outputs in the highly specialized field of drug development, the participation of biologists with expertise in both drug discovery and AI is essential.
FRONTEO Drug Discovery AI Factory's team of researchers consists of biologists with extensive knowledge in the life science field, who make full use of our self-developed AI engine and applications. By combining a deep understanding of the needs of pharmaceutical companies with proprietary analytical methods, we are able to present targets that have not yet been reported and have a high probability of success, and to generate its hypotheses.
Achievements - Include the Case of in Vivo Confirmation of Drug Efficacy
We have already been entrusted with more than 20 projects to meet a variety of client needs, including target discovery, indication discovery, drug combination, and biomarker discovery. In one case, we proposed 20 highly novel target molecules for which no direct link to the disease was described in the paper, and five of them were confirmed to work in vitro, and one was confirmed to work in vivo. We respond to the needs of pharmaceutical companies, such as the launch of new disease areas, pipeline enhancement, early expansion of indications, and the re-utilization of candidate compounds, by combining AI and biologists who make full use of it.
- Hypothesis Generation AI as Part of the Drug Discovery Process
Until now, hypotheses for new drug development have been assembled over time through the individual efforts of researchers. FRONTEO supports drug discovery research for pharmaceutical companies by renewing the "hypothesis generation" process, which is the beginning of the drug discovery process, utilizing AI to generate hypotheses dramatically faster than ever before and continuously.
FRONTEO's Drug Discovery AI Factory, as a platform specializing in hypothesis generation, will incorporate AI into the future drug discovery process and further promote social implementation of AI in the field of drug discovery.