Cintecabio AI New Drug Platform and Bio-IT-World's Launchpad Program.

Cintecabio AI New Drug Platform and Bio-IT-World's Launchpad Program.

1. Introduce companies that promote research on new drug candidates
   - Amelia Crisselle, PhD, Senior Vice President of Business Development at Syntekabio.
   - On the last day of the conference, an introduction to Cintecabio *trying to deliver new drug candidates to patients faster.
   - We will enter the presentation, emphasizing that we will introduce and confidently explain the efficient solutions provided by Cintecabio.

2. Introducing AI drug discovery companies in ️
   - Syntekabio, an AI drug discovery company, is headquartered in South Korea and was founded 15 years ago as a supercomputing center for ETRI and big data.
   - Over time, we then developed *computational methodology*, *structural biology*, AI and ML, and three-dimensional molecular spaces.
   - Operate an in silico task and a *global zero network* with a supercomputing center specializing in AI and medicine that opened last year to check all predictions with *real experimental results*.
   - Emphasizing that drug discovery is a complex and time-consuming process, it combines expertise and AI solutions in *preclinical fields* to reach targets to IND packages in two years.

3. Search for new drug candidates and discover early drugs using AI
   - We are trying to provide patients with the best drugs as soon as possible and we are detecting a great demand in the medical field.
   - Measures need to be taken for patients who are unable to receive nursing care or who do not have a valid drug to help with a particular condition or problem.
   - To this end, the goal is to provide high-speed computing systems and know-how, and to simplify drug discovery in the pre-clinical period using AI and deep learning.
   - It focuses on improving the speed and precision of early-stage drug discovery by adopting Syntekabio's DeepMatcher platform as a core technology.

4. Inspection of the molecular chemical space with an AI bio supercomputer.
   - The task of permuting the molecular chemical space against the protein target required a lot of computational work.
   - I have built my own supercomputing center to do this with the rigor required by my system. This includes ABS, an AI bio supercomputer, and there are 10,000 servers in operation.
   - The operating system is designed exclusively for these calculations, and everything is optimized for *precision* and *efficiency*.
   - By the end of 2026, two additional buildings will be used online to expand server capacity, and the model is learning to do better on this prediction task over time.

5. How much is required for model training and forecasting
   - We need to secure and model an appropriate dataset for prediction.
   - You must find and evaluate a model that is appropriate for that problem or prediction.
   - Thereafter, there is a model training phase, which often relies on data.
   - For example, we used a physics-based approach, but it could be a different dataset. We need to make a lot of decisions about where to train our model.
   - Finally, the results obtained after running the predictions are as good as the results in vitro or physiology. To confirm this, we need to conduct synthetic experiments from outside.

6. Leverage *cloud solutions* to improve model training and forecasting
   - After validation of the model, we complement the model according to the failure of the test and want to proceed through *experimental animal testing* and IND steps.
   - Leveraging cloud solutions *automates connections between models and chemicals*, aiming for precise prediction and results.
   - In silico work communicates with chemical work, *improves the model*, and *extracts precise results*.
   - This allows for more accurate results.

7. Improving experimental results with AI in drug development
   - The solution is physically trained and applicable to a wide range of therapeutic fields, and can address different proteins and peptides to meet different drug discovery needs.
   - Experimental validation and testing are used to improve the model over time, incorporating chemical experiments, in vitro experiments, and in vivo experiments into AI discovery processes to minimize failures early.
   - In practice, experimental verification and testing are used to improve experimental results, and optimization processes exist for checkpoints around chemistry and synthesis.
   - Using Global C Partners and synthetic partners together, we don't have to wait for laboratory testing, and we continue to optimize it to improve how predictable the model can be over time.

8. Describe how the protein heat generation algorithm works
   - *Detailed on the hit generation algorithm Auto Binding Pose*.
   - A method of *creating hit candidates* using the three-dimensional molecular structure of the target protein is introduced.
   - After mapping the protein sacs and creating a frame to fit in the sacs, the candidate molecules are placed in space to find the best direction.
   - These candidates are stored in a virtual library and evaluated by 3D CNN scoring, *minimizing collisions and optimizing getting into the most suitable positions*.

9. Select the compounds required by ️ molecular power simulation
   - First, we select about the top 10,000 and look at clusters using CNN with excellent performance.
   - Through molecular power simulations, we select the top 100 compounds for optimal coupling and lowest tensile.
   - As a result, the introduced "Autobinding force algorithm" finally exceeds the industrial standard and selects the best compound.
   - Through this, the selected compound proceeds to the next step through the addition test, and at this time, efforts are made for good bonding.

10. Lead Optimization and Development of Leading New Drugs through ️ Molecular Dynamics
   - *Outperforms the industry's best by 1%* at the precision level measured by the auto-docking algorithm, resulting in reliable hits.
   - After testing on the top 100 hits *invitro*, we move on to the existing drug lead, the *Algo* process.
   - Using chemical Scarfolds and R groups, we test 1 to 3 million mean *lead structures* in the computational space in combination with 20,000 R groups.
   - To find up to 10,000 compounds, run CNN scoring and molecular dynamics simulations, and finally conduct a medical compound test with *intensive resolution* to select the final 20 compounds and transfer them to the laboratory.

11. Compare RMSD performance for protein-ligand complexes and share test results
   - The performance of Synca Bio was dark blue, and the industry standard detected that the RMSD levels of protein-peptide complexes, which appeared light blue, were significantly lower than the industry standard.
   - The graph on the right shows how low RMSD can be detected with one algorithm model, targeting protein ligand complexes.
   - Looking at the IC50 plot of the reads for clk2, most of the several leads candidates that emerged as algorithmic models after treatment of the initial compounds improved their binding to clk2, with 32x improvements, such as achieving an IC50 of 4 nanometers.
   - This took about 5 months to review, with very good results.

12. A Study on Target FLT3 and Activity Improvement for Ido and tdo
   - In an activity-enhancing study targeting FLT3, six of the seven compounds outperformed the original compound in binding to the target.
   - In a dual-target experiment with Ido and tdo, studies have been conducted to improve compounds for two different immune checkpoint inhibitors.
   - In the second experiment, the top compound was enhanced about 14-fold over a 12-month period relative to the original compound.
   - It has been actively studied for two targets, improving the binding site and activity.

13.Find collaborative partners for drug discovery and clinical trials
   - We were able to reduce the timeline by finding a rapid synthetic partner that provided the compound in a small amount of time.
   - Syntekabio Launchpad is a product and platform that leverages cloud-based infrastructure and high-speed computing centers to perform drug discovery tasks.
   - We have established a parallel pipeline for drug discovery and are conducting research simultaneously on multiple treatment areas and multiple targets, covering serious unmet medical fields.
   - We are currently working to invest in these candidate drugs, take risks, and accelerate them into clinics, and hope to find collaborative partners and collaborators with expertise.

14. Syntekabio's expertise in accelerating discovery of new drugs with AI.
   - The workflow continues, and checks and optimizes the results in real time at the intermediate stage of setting new goals and achieving them.
   - Syntekabio accelerates *risk reduction* and *test* of new drug candidates through its AI platform, helping clients quickly expand their assets and build pipelines.
   - It is a disease-neutral AI platform that operates flexibly by reflecting *customer requirements* and provides previously risk-reduced assets when desired, as well as searching for other goals or disease indicators to find new candidates.
   - We highlight that we are accelerating our new drug development efforts through investments in technology and workflow, and we can refer you to an expert if you have technical questions with an expert from Syntekabio.
   - Emilia introduced Syntekabio's AI expertise and TurnKey assets, gave us a glimpse of the potential for future development, and if you have any technical inquiries at booth 22, a conversation window, you can refer experts to you.