Drug Target Identification and Omic Methods

Drug Target Identification

• Drug target identification finds disease-related, druggable targets (proteins & nucleotides).
• Key technologies include genomics, transcriptomics, proteomics, metabolomics, and 3D structures.
• Target identification combines in silico (computer-based) and laboratory techniques.

Omic-Based Methods

• Omic methods offer a multi-layered understanding of molecular biology.

• Genomics*

• Genome-wide association studies (GWAS):

  • Examine genomes to link genetic markers to diseases.
  • Steps include data collection, genotyping, quality control, analysis, and validation.
  • Helps identify risk factors for diseases like diabetes and cancer.
  • Challenges include handling rare variants and multifaceted traits.

• Variant Effect Prediction:

  • Predicts disease impacts of genetic variants, valuable for rare/novel changes.
  • 2 AI approaches:
    • Using known variant effects (e.g., PrimateAI, leveraging evolutionary conservation, and protein structure prediction).
    • Inferring effects from protein evolutionary landscape (e.g., protein language models, using patterns in protein sequences and log-likelihood ratios).

• Transcriptomics*

• RNA sequencing (RNA-Seq):

  • Bulk RNA-Seq: Analyzes gene expression in a cell mixture, comparing, for example, pre and post-drug treatment.
  • Single-cell RNA-Seq (scRNA-Seq): Profiles individual cell gene expression, allowing for identifying cell types (heterogeneity).
  • Spatial transcriptomics: Measures gene expression in tissues preserving spatial information, aiding in cell-cell interactions.

• Proteomics*

• Large-scale study of proteins (proteome) expressed in specific conditions.

• Metabolomics*

• Study of metabolites (small molecules) in cellular metabolism.

• Integration of Omic Approaches:*

• Combining multiple omic methods creates a more comprehensive view of complex disease processes.

Molecular Structures in Drug Discovery

• Protein 3D structures are vital for drug design.
• Structure-based drug design uses 3D structures to design drug molecules that bind specifically to targets.
• Challenges of Targeting Proteins:*
• Some proteins lack discernible binding sites for ligands, making them difficult to target with drugs.
• Determining whether a protein is a viable drug target.

Protein Structures

• Levels of protein structure:

  • Primary: Amino acid sequence
  • Secondary: Local motifs (alpha helices, beta sheets)
  • Tertiary: Overall 3D structure of one protein chain
  • Quaternary: Interactions of multiple subunits forming a complex.

• Protein Data Bank (PDB): Global repository of protein, nucleic acid, and complex structures.

Structure Determination Methods

• 3 Main techniques:
  1. X-ray crystallography: High resolution, widely used, but challenging crystal preparation, & static view.
  2. NMR spectroscopy: High resolution, studies proteins in solution (natural state), challenging sample prep, and small proteins.
  3. Cryo-EM: Easy sample prep, studies proteins in their native state, small sample size, good for large complexes but lower resolution, expensive equipment

Protein Structure Prediction

• Need for prediction: Experimental structure determination is time-consuming and expensive.

• Prediction methods (2 main categories):

  1. Template-based modeling:
     • Homology modeling: Leverages high sequence similarity.
     • Protein threading: Low sequence similarity – aligns sequence to existing structures.
  2. Ab initio modeling: Predicts structure from sequence alone without comparing to known structures, but less accurate.

AlphaFold2

• AI system for groundbreaking accuracy in protein structure prediction, developed by Google DeepMind.
• Input: Protein sequence
• Output: Detailed 3D structure (atoms, backbone, side chains)
• Uses multiple sequence alignments, and neural networks.
• High accuracy (e.g. high TM-scores and GDT-scores), rivals experimental structures.
• Revolutionizes drug discovery, structural biology, and molecular biology.

Resources for Protein Structure Prediction

• Databases (AlphaFold Protein Structure Database, ESM Metagenomic Atlas)
• Online tools (ColabFold, ESM Fold Sequence)

Transition to Drug Screening and Design

• Understanding protein structure is crucial for structure-based drug discovery leading to hit identification and lead optimization.