New simulation framework DKOsim advances CRISPR-based gene to gene interaction research

MD Anderson Research News April 30, 2026

• Double-CRISPR Knockout Simulation (DKOsim) is a computer-based “virtual lab” that simulates genomic experiments 
• Studying how genes interact in real cells is difficult, expensive and often produces uncertain results
• DKOsim allows researchers to test ideas and methods using realistic controlled data while designing more reliable real-world experiments

Researchers at The University of Texas MD Anderson Cancer Center developed Double-CRISPR Knockout Simulation (DKOsim), a novel computational framework designed to address longstanding challenges in profiling gene to gene interactions. 

“DKOsim provides a much-needed bridge between experimental and computational biology,” said John Paul Shen, M.D., assistant professor of Gastrointestinal Medical Oncology, “It enables scientists to systematically test a hypothesis, optimize experimental design and benchmark analytical tools under controlled conditions.”

The study, published in PLOS Computational Biology, was led by Shen and first author Yue Gu, Ph.D., a recent doctoral graduate from the Shen Lab. The findings may help streamline genetic interactions detection through optimized computational methods and guide the development of next-generation dual-knockout CRISPR screens.  

Why is DKOsim important?

Genetic interactions play a crucial role in understanding complex biological systems and diseases, yet they remain difficult to measure accurately. Traditional laboratory approaches, such as multiplexed CRISPR screening, are time intensive, costly and often limited by technical challenges. Meanwhile, computational methods struggle due to the lack of reliable databases, resulting in inconsistent findings across experiments.

To address these challenges, researchers developed DKOsim, a Monte Carlo-based simulation strategy that models both single and double gene knockout experiments while accounting for factors like cell growth dynamics and how well Guide RNAs perform. The framework generates realistic synthetic datasets with known interaction values, enabling rigorous evaluation of analytical methods.

The framework has demonstrated strong alignment with real laboratory data, accurately reproducing observed patterns in genetic interactions scores. 

***

Funding was supported by the Col. Daniel Connelly Memorial Fund, the Andrew Sabin Family Fellowship Award, the Cancer Prevention & Research Institute of Texas (RR180035 & RP240392), the Appendix Cancer Pseudomyxoma Peritonei Research Foundation and a Conquer Cancer Career Development Award. A full list of collaborating authors and their disclosures can be found with the full paper in PLOS Computational Biology.