PriFold: Biological Priors Improve RNA Secondary Structure Predictions
Authors: Chenchen Yang, Hao Wu, Tao Shen, Kai Zou, Siqi Sun
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments demonstrate that Pri Fold achieves state-of-the-art (SOTA) results in RNA secondary structure prediction on benchmark datasets such as bp RNA, RNAStr Align and Archive II. These results not only validate our prediction approach but also highlight the potential of integrating biological priors, such as global characteristics and evolutionary information, into RNA structure prediction tasks, opening new avenues for research in RNA biology and bioinformatics. |
| Researcher Affiliation | Collaboration | 1Research Institute of Intelligent Complex Systems, Fudan University 2Shanghai Artificial Intelligence Laboratory 3Zelixir Biotech 4Net Mind.AI EMAIL, EMAIL |
| Pseudocode | No | No explicit pseudocode or algorithm blocks are present in the main text. The methodology is described narratively and visually in Figure 2. |
| Open Source Code | Yes | Code https://github.com/BEAM-Labs/Pri Fold |
| Open Datasets | Yes | Our study utilizes three benchmark datasets to evaluate the performance of our model. To establish a standard benchmark for comparison, we use RNAStr Align (Tan et al. 2017) as our primary dataset. ... we include Archive II (Mathews 2019), a widely used testing dataset... we incorporate bp RNA (Danaee et al. 2018) dataset... |
| Dataset Splits | Yes | We split RNAStr Align into training set and test set following UFold, E2Efold, and MXFold2. ... We divided bp RNA into training, validation and test set following SPOT-RNA. |
| Hardware Specification | No | No specific hardware details (such as GPU/CPU models or memory) are provided in the paper for running the experiments. |
| Software Dependencies | No | The paper does not explicitly state specific software dependencies with version numbers, such as Python versions, deep learning framework versions (e.g., PyTorch, TensorFlow), or other library versions. |
| Experiment Setup | Yes | λ is a hyperparameter that scales the probability, allowing adjustments to the influence of the probability term on the final matrix value. ... The model achieves the highest F1 score of 0.737 with λ = 0.01, providing the best balance between precision and recall. Based on these observations, we selected λ = 0.01 for all subsequent experiments. ... We systematically vary α from 10% to 100% in 10% increments and evaluate the model s performance of each setting. ... Notably, the most significant gains are observed with a replacement ratio of 30%... |