ARK : Multi-Strategy "Combinatorial Drug Design" platform

World’s first combinatory drug development platform through integrated analysis

  • Gene expression pattern

  • 3D structures of protein/chemicals

  • Literature analysis on Medicine and Pharmacy

AI Target Finder - Drug target prediction

Selected as ICT project of 2017 by Ministry of Science, ICT and Future Planning Technology that predicts protein, which will become the drug target, from the patient genome data

Link to the article link
  • Published on Bioinformatics in October 2018. Patent registered.
  • Requires large set of patient genome data
  • Can be applied to diseases with large amount of accumulated data, such as cancer and immunological diseases

VLab[Compound Library] – Single drug structure prediction

CombiNet – Designing drug combination

  • Only the gene expression data - excluding the information on drug structure or effect – are used.
  • The gene information of patient is connected with NOTE-R and analyzed as an important pathological network.
  • Comparison of the gene expression patterns.
  • The existing methods only includes analysis of the genes with significant changes
  • The key characteristic of CombiNet is the machine learning analysis technology, which discovers characteristic pathological networks in patients through literature analysis and integrated gene network analysis
  • It has an advantage of accurate prediction of drug effects through comparison analysis on the key genes from the pathological network.
  • Single drugs are ranked by their effects for each pathological network.
  • Synergetic effects and drug-drug interaction with other drugs are considered.
  • All networks are effectively controlled to develop the optimal combinatory drug.
  • Has a separate system to predict risks of DDI

VLab[Protein Library] – Discovering off-target

  • Unlike docking, this is a reverse screening technology that is used to discover a new off-target of drugs.
  • I can be utilized to find new drug effects or to predict side-effects or toxic properties.
  • The same basic energy calculation method as in the virtual compound screening method is used, but the dataset used for learning is different therefrom. It requires more variety of refined 3D structure information of protein.

NeuroRG - Neuronal cell image-based drug efficacy assessment

  • Cell morphology-based medicinal effect evaluate
  • HTS equipment enables simultaneous analysis of large quantities of drugs
  • Analyze various cells such as Neuron, Astrocyte, Oligodendrocyte, Microglia, and Myocyte
  • Image analysis of the pathological condition
  • Due to the nervous system connected by a complex network, a single target drug’s effectiveness is often insufficient.
  • It is more effective to evaluate a drug's efficacy with pathological conditions representing the disease instead of a single target.