Building, benchmarking, and exploring perturbative maps of transcriptional and morphological data

Area of Science:

• Systems Biology
• Genomics
• Computational Biology

Background:

• Advancements in genetic perturbation technologies (e.g., CRISPR-Cas9) and high-dimensional assays (e.g., RNA-sequencing, microscopy) enable genome-scale reverse-genetics studies.
• These studies generate large datasets that can be integrated into unified
• maps of biology
• for comprehensive analysis.
• Constructing these maps involves complex experimental and computational choices, necessitating robust evaluation methods.

Purpose of the Study:

• To establish standardized terminology and benchmark procedures for assessing the quality of perturbative maps of biology.
• To evaluate the performance of different map construction strategies across various datasets and experimental conditions.
• To demonstrate the utility of these maps for discovering novel biological insights, including gene functions.

Main Methods:

• Standardized terminology for perturbative map building was established.
• Key benchmark classes were introduced to systematically assess map quality.
• Eighteen genome-scale maps were constructed from four diverse datasets using different perturbation technologies and readout modalities.
• Benchmark metrics were generated to analyze map performance and identify factors influencing variations.

Main Results:

• The study established a framework for evaluating maps of biology, enabling systematic and unbiased quality assessment.
• Performance variations across the 18 constructed maps were investigated, providing insights into optimal construction strategies.
• The utility of these maps was demonstrated through the successful suggestion of functional roles for two uncharacterized genes.

Conclusions:

• Standardized benchmarks are crucial for the reliable construction and interpretation of maps of biology.
• These maps serve as powerful tools for integrating diverse biological data and uncovering complex cellular relationships.
• Maps of biology facilitate hypothesis generation and accelerate the discovery of gene function and biological mechanisms.