Concepts

Define key terms used across design, execution, and review.

Omics Foundation Model

An omics foundation model is a generative AI that produces omics data (e.g., gene or protein expression) under specified conditions. It enables rapid exploration of hypotheses before conducting physical experiments.

Condition-driven Generation

Outputs are determined by combinations of conditions such as disease context, cell type, timepoint, genotype, and treatment or culture parameters—not just a single target or label.

Common Generation Scenarios

• knockdown: Simulate suppression of a gene or protein
• Time-series generation: Create intermediate or future timepoints
• Treatment/culture simulation: Model drug or environmental conditions

Baseline vs Condition

The baseline represents the reference state, while conditions define the scenario to be generated (e.g., control vs knockdown, healthy vs disease).

Output Interpretation

• Marker changes: direction and magnitude
• Condition separation: distinguishability across scenarios
• Candidate prioritization: effect size, consistency, biological plausibility
• Downstream use: validation design, ranking, reporting

Pseudo-blotting & Candidate Review

Lite.KnockG provides quick inspection of marker changes via pseudo-blotting. Advanced versions support model training, time-series analysis, multi-target simulation, and integrated literature/patent/network analysis.

In silico vs Wet-lab

KnockG complements wet-lab experiments by enabling broader exploration of conditions and narrowing down candidates before physical validation.