日本語 / English

~最新のScience Technologyを有する英米3社の特別講義・CBI学会スポンサー企業企画~


『次世代GPUシミュレーションと iPS疾患モデルが融合する創薬イノベーション』

Innovative drug discovery platform combining next generation GPU simulation and disease modelled iPS cells.



Abstract - 次世代GPUシミュレーションと iPS疾患モデルが融合する創薬イノベーション
Biospire-Lifematics CBI seminar A4x2 abs
PDFファイル 1.8 MB






Dr Giovanna Tedesco (Cresset)

Dr Clinton Threlfall (SilcsBio)

Dr Marcus Yeo (DefiniGEN)


主催 Biospire株式会社、ライフマティックス株式会社



座長 慶應義塾大学薬学部 池田和由 氏






米SilcsBio社からは、小分子プローブを導入したマイクロ秒アンサンブル MD法をコア技術にしたヒットtoリードにおける創薬探索研究の事例をご紹介いたします。これからの計算化学において、必要不可欠なGPU高速化シミュレーションの先駆けとしてリアルタイムMDの最新技術、及びPharmacoporeの作成、ポケット発見、バーチャルスクリーニングの手法についてフォーカス致します。




Predicting biological activity using the electrostatic complementarity of protein-ligand complexes


Matthias Bauer, Mark Mackey, Giovanna Tedesco


Electrostatic interactions between small molecules and their respective receptors are a key contributor to the free energy of binding. Assessing the electrostatic match between ligands and binding pockets provides therefore important insights into why ligands bind and what could be changed to improve binding. The polarizable XED force field is an excellent base for calculating electrostatic properties due to its description of anisotropic atomic charge distributions and relatively modest computational costs. By computing electrostatic potentials for both ligand and protein with XED, the Electrostatic Complementarity™ of complexes can be assessed via (1) inverse correlation of the respective local electrostatic potentials (Pearson or Spearman rho rank tests) or (2) calculating a normalized surface complementarity integral, yielding electrostatic complementarity scores. The latter approach also allows visualization of the local electrostatic matching on the van der Waals surface to identify electrostatic clashes and inform ligand design. We present the theoretical background of our electrostatic complementarity descriptors along with several case studies showing the practical application of the scores to the prediction of activity and of the visualization to ligand design.



Site Identification by Ligand Competitive Saturation (SILCS) reproduces experimental binding trends for 31 TrmD ligands


Sirish Kaushik Lakkaraju, Clinton Threlfall, Olgun Guvench, Alexander D. MacKerell Jr


Site-Identification by Ligand Competitive Saturation (SILCS) computational functional group mapping provides insights into the binding preferences of a target protein that can be used qualitatively and quantitatively to drive ligand design. SILCS is a robust structure-based approach that gives information-rich Grid Free Energy (GFE) FragMaps that encompass critical aspects such as protein flexibility and explicit solvation. Here we describe the use of the SILCS approach on tRNA methyltransferase (TrmD) and 31 ligands belonging to two series originally donated by GSK and made publicly-available through Community Structure-Activity Resource (CSAR) and the D3R Database.

SILCS-MC sampling of ligands in the field of the FragMaps yields Ligand Grid Free Energy (LGFE) scores. SILCS scoring correctly predicts favorable vs. unfavorable modifications relative to a reference ligand (27/30 predictions correct). Additionally, SILCS FragMaps recapitulate functional group patterns of both series of ligands. This information can be used to drive design and optimization visually.


Validating novel determinants of metabolic disease using an integrated CRISPR/Cas9 iPS differentiation platform technology approach


Marcus Yeo, Filipa Soares, Masashi Matsunaga, Ludovic Vallier


To validate the new therapeutic target candidate outputs from disruptive AI approaches and a range “omic” technologies: genetic, epigenetic and transcriptomic, proteomic, and metabolomic, next generation phenotypic screening platforms are required. DefiniGEN platform combines CRISPR/Cas9-gene editing, Nobel prize winning Yamanaka iPS stem cell and GMP-compatible directed differentiation to generate predictive disease models for metabolic disease which are reflective of the in vivo state. The platform can generate models for complex diseases such as NAFLD/NASH in which disease implicated loci are systematically modified alongside isogenic control lines which are identical to the disease variant apart from the introduced mutation. This enables us to determine the contribution of  a particular gene variant to the disease state. This approach is particularly powerful for complex diseases where multiple genes and variants impact on the severity of disease progression.