Menu

Drug Discovery Platform with Hardware

Analogue Platform

Case Study

OUR OFFERINGS

Case Study

Small Molecule Design For Neuronal Disorder

The client approached us with the objective to optimize their drug discovery process, aiming to reduce costs and accelerate the timeline for bringing new drugs to the advanced phase. Our team proposed the integration of in silico AI-assisted drug design solution to complement experimental methods and enhance the overall drug development pipeline.

Know More

Peptide Design to Inhibit Protein Aggregation

A protein ‘X’ forms an aggregate and the deposition of these aggregates results in a disorder. client Biotech was looking for a peptide/oregano peptide that can bind at the interface of the oligomer/dimer and inhibit the initiation of the aggregation process.

Know More

Identifying & Targeting new Target for Infectious Diseases

Client’s objective was expediting drug development and requiring drug targets for infectious disease ‘A’, so we proposed a network-based drug target identification approach. Leveraging in silico techniques and approved drugs, our innovative strategy aims to significantly reduce the time and cost associated with the process. This approach holds promise for rapidly identifying potential treatments, accelerating the quest for effective therapies against disease ‘A’.

Know More

Hit Molecule Identification in Neuronal Disorder

INTRODUCTION

The client approached us with the objective to optimize their drug discovery process, aiming to reduce costs and accelerate the timeline for bringing new drugs to the advanced phase. Our team proposed the integration of in silico AI-assisted drug design solution to complement experimental methods and enhance the overall drug development pipeline.

OBJECTIVES

The traditional drug discovery process is a lengthy and expensive endeavor, often taking years and significant financial investment. Client faced several challenges in its existing process:

  • Methods for swift identification of hit compounds targeting 9 specific drug targets they have pinpointed.
  • Establishing a disease-specific compound library for their research focused on the identified drug targets.
  • Techniques for identifying compounds with desired physico-chemical properties that align with their research criteria.

SOLUTION

To address client’s challenges, our team proposed integrating in silico drug design techniques complemented with AI/ML-based prediction into their drug discovery workflow.

  • Identifying the correct binding site: Each protein has been predicted with 2-3 top binding sites, using bioinformatics tools we identified the best site to target that is guided by the functionality of the protein.
  • Detecting the Druggability of the identified binding site: Thermal stability pattern of the binding site detected to approve or reject the protein as “Drug Target”. Accordingly, 2 proteins dropped from the list.
  • Creating a Library: With given conditions by the client, a compound library of 1003 compounds.
  • Scoring Function Design: A scoring function designed based on the Boltzmann distribution.
  • Hit Selection on Scoring Function: 50 best hits selected for every 7 targets on the scoring function.
  • AI Model building: A graph neural network model was built to detect the IC50 of these 50 hits with their corresponding target.
  • Selecting 10 compounds of IC50: AI model was deployed and the top 10 IC50 compounds were identified. The predicted IC50s (in nM) of one protein are:

3.254, 3.263, 3.288, 3.291, 3.291, 3.291, 3.318, 3.322, 3.34, 3.354

Experimental Testing the compounds: The client tested 4 compounds for the most promising protein target and all 4 showed biological activity while 2 of them showed better biological activity on the cell line compared to the market standard.

BENEFITS

  • Reduced Timeline: The drug discovery timeline was significantly reduced by up to 30-40%, as virtual screening narrowed down the pool of potential candidates.

Cost Savings: The number of compounds tested was substantially reduced, leading to cost savings of approximately 50-60% in the early hit identification

Hit Molecule Identification in Neuronal Disorder

OBJECTIVES

A protein ‘X’ forms an aggregate and the deposition of these aggregates results in a disorder. client Biotech was looking for a peptide/oregano peptide that can bind at the interface of the oligomer/dimer and inhibit the initiation of the aggregation process.

Solutions

Growdea suggested to client Biotech to perform a detailed computational study to understand the interaction interface of the protein dimer using computational protocol and design multiple peptides based on the template. Later, a machine learning approach could be deployed to modify the peptide and predict the binding affinity.

PROCESS FOLLOWED

Train an AI model to detect the possible interface of protein

In first step, we developed an AI algorithm that can detect the most probable stretch of the protein that could be the interface of the dimer. It detected 11 interface prone for protein-protein interaction.

Perform Protein-Protein Docking

The AI detected interface used to perform the protein-protein docking and 100 poses are generated for each pair. Total poses =1100.

Build AI model to predict the Kd

The AI-detected interface is used to perform the protein-protein docking and multiple poses are generated. The data had Kd in Molar (M) concentration.

Design a peptide based on the interface template

In the next step, the AI model was used to design the best peptide that can be used to compete with the protein monomers and stop to form a dimer. Based on the Kd prediction 6 peptides given to the client with Kd: 3.4 * 10-9 M, 2.9*10-9 M, 1.1*10-9 M, 1.4.0*10-9 M, 4.0*10-9 and 2.86*10-9 M,

Fluorescence Assay

Here, 2 peptides were found to interact with the protein, shown in the fluorescence assay.

Concentration of Px(protein): Peptide = 1:1

No major interaction detected at 1:1 concentration
Concentration of Px(protein): Peptide = 1:10

This results showed that p3 and p4 interacts significantly with the protein at 1:10 concentration.

BENEFITS

  • Speed and Efficiency: We delivered the entire project in 4-5 months of time, saving time compared to traditional trial-and-error methods
  • Precise Peptide Design: AI algorithms can optimize peptide sequences for desired properties, such as improved binding affinity
  • Reduced Costs: By streamlining the design and testing process, AI can help reduce the costs associated with developing therapeutic peptides. Only 6 peptides testing.

Identifying & Targeting new Target for Infectious Diseases

INTRODUCTION

Client’s objective was expediting drug development and requiring drug targets for infectious disease ‘A’, so we proposed a network-based drug target identification approach. Leveraging in silico techniques and approved drugs, our innovative strategy aims to significantly reduce the time and cost associated with the process. This approach holds promise for rapidly identifying potential treatments, accelerating the quest for effective therapies against disease ‘A’.

OBJECTIVE

The client seeks to accelerate the drug development process. However, traditional drug discovery methods are known for their high costs and time-consuming nature.

  • A network-based approach for identifying potential drugs targeted toward the treatment of disease ‘A’ infection.
  • The methodology integrates protein-protein interactions (PPI), host-disease interaction proteins, drug-target networks, and pathway analysis to facilitate the identification of potential drug candidates.
  • The study uses a network-based approach to reduce costs and shorten timelines for drug discovery.

SOLUTION

  • Protein-protein interaction (PPI) data: We collected experimentally verified and predicted PPI data from various databases, including STRING and BioGRID.
  • Gene expression data: Transcriptomic data from disease-infected cells were obtained from public repositories.
  • Network Construction: Integration of PPI data: The collected PPI data were integrated to construct a comprehensive disease protein interaction network.
  • Network analysis: Topological properties, such as centrality and connectivity, were calculated to identify highly influential proteins within the network.
  • Functional enrichment analysis: Gene ontology and pathway enrichment analyses were performed to understand the biological functions and pathways associated with the identified proteins.
  • Prioritization of drug targets: Using network-based algorithms, such as random walk and diffusion-based methods, candidate proteins were prioritized based on their potential impact on the network.

BENEFITS

Accelerated Results: The method’s expedited process ensures quicker drug development, addressing the client’s need for rapid outcomes.

Cost Efficiency: By bypassing traditional resource-intensive steps, this method proves cost-effective, aligning with the client’s goal of minimizing expenses.

Streamlined Process: The approach’s efficiency reduces time consumption, allowing the client to achieve their objectives promptly while optimizing resource allocation.