Related Experiment Video
Updated: Jul 19, 2025

Author Spotlight: Unlocking New Insights in fNIRS Studies - A Novel Framework for Inter-Brain Synchrony Analysis
Published on: October 6, 2023
Inferring couplings in networks across order-disorder phase transitions
Vudtiwat Ngampruetikorn1, Vedant Sachdeva2, Johanna Torrence2
1Initiative for the Theoretical Sciences, The Graduate Center, CUNY, New York, New York 10016, USA.
Direct coupling analysis (DCA) effectively infers interactions from Ising models, but its performance depends on data distributions. Optimal accuracy occurs at intermediate temperatures, with simpler methods outperforming DCA when data is limited.
Area of Science:
- Statistical physics
- Computational biology
- Machine learning
Background:
- Statistical inference is crucial in science but not fully understood.
- Understanding the interplay between models, inference, and data structure is key.
- Direct Coupling Analysis (DCA) is a successful method for amino acid sequence data.
Purpose of the Study:
- To characterize DCA's efficacy in inferring pairwise interactions.
- To analyze Ising models on random graphs using DCA.
- To explore data regimes and phase transitions relevant to inference.
Main Methods:
- Simulating ferromagnetic Ising models on random graphs.
- Applying Direct Coupling Analysis (DCA) to infer pairwise interactions.
- Analyzing inference quality across different data-generating distributions and temperatures.
Main Results:
- Inference quality is highly dependent on data-generating distributions.
- Optimal accuracy is achieved at intermediate temperatures, balancing order and noise.
- DCA is outperformed by simpler correlation thresholding methods with limited data.
Conclusions:
- DCA's success is linked to specific data regimes, particularly at low temperatures.
- The study provides insights into the operational limits and strengths of DCA.
- Findings highlight the broader interaction between statistical inference and data structure.
More Related Videos
08:55Methods of Ex Situ and In Situ Investigations of Structural Transformations: The Case of Crystallization of Metallic Glasses
Published on: June 7, 2018
10:44Inherent Dynamics Visualizer, an Interactive Application for Evaluating and Visualizing Outputs from a Gene Regulatory Network Inference Pipeline
Published on: December 7, 2021
Related Concept Videos
Phase Transitions
Phase Diagrams
Phase Diagram
¹H NMR: Interpreting Distorted and Overlapping Signals
As Δν decreases and the signals move closer, the doublets appear increasingly distorted. The intensities of the inner lines increase at the cost of those of the outer lines as the signals are...
Phase Transitions: Sublimation and Deposition
Phase Transitions: Melting and Freezing