# SidechainNet

> a protein structure prediction dataset

**Wikidata**: [Q114878822](https://www.wikidata.org/wiki/Q114878822)  
**Source**: https://4ort.xyz/entity/sidechainnet

## Summary
SidechainNet is a protein structure prediction dataset that serves as both a biological database and open-source software. It provides tools and data for predicting protein structures, with the latest stable version released in November 2023.

## Key Facts
- SidechainNet is a protein structure prediction dataset classified as both a biological database and open-source software
- It was authored by David Ryan Koes under the 3-clause BSD License
- The current stable version is 1.0.1, released on November 2, 2023
- It's built upon ProteinNet and utilizes PyTorch as its machine learning framework
- The dataset is provided in Python Pickle serialized data (v4) format
- The source code repository is available at https://github.com/jonathanking/sidechainnet
- It was published in the journal Proteins
- It cites works such as "Structure prediction for CASP8 with all-atom refinement using Rosetta" and "Improved protein structure prediction using predicted interresidue orientations"

## FAQs
### Q: What is SidechainNet?
A: SidechainNet is a protein structure prediction dataset that serves as both a biological database and open-source software. It provides tools and data for predicting protein structures, which is essential for understanding biological functions and drug development.

### Q: Who created SidechainNet and under what license is it distributed?
A: SidechainNet was created by David Ryan Koes and is distributed under the 3-clause BSD License, which allows for free use and redistribution with permissive access to the source code.

### Q: What technologies is SidechainNet built upon?
A: SidechainNet is built upon ProteinNet and PyTorch. It utilizes the machine learning capabilities of PyTorch and the foundational data from ProteinNet to provide its protein structure prediction functionality.

### Q: How can I access SidechainNet?
A: SidechainNet is available through its GitHub repository at https://github.com/jonathanking/sidechainnet. The dataset is provided in Python Pickle serialized data (v4) format, making it compatible with Python-based machine learning workflows.

### Q: What research does SidechainNet cite?
A: SidechainNet cites several important works in protein structure prediction, including "Structure prediction for CASP8 with all-atom refinement using Rosetta" and "Improved protein structure prediction using predicted interresidue orientations," reflecting its foundation in established research methodologies.

## Why It Matters
SidechainNet represents a significant resource in the field of computational biology and protein structure prediction. By providing a comprehensive dataset and open-source tools, it enables researchers to develop and test new algorithms for determining protein structures—a critical step in understanding biological functions and designing therapeutics. Its integration with PyTorch makes it accessible to the machine learning community, fostering innovation in this challenging area. The dataset's development and continued updates ensure that researchers have access to current, high-quality data for training and validation of their models, accelerating progress in this fundamental scientific problem.

## Notable For
- Being a comprehensive protein structure prediction dataset that combines structural data with machine learning frameworks
- Its integration with PyTorch, making it accessible to both computational biologists and machine learning researchers
- Being built upon the established ProteinNet foundation while adding side chain prediction capabilities
- The active maintenance and regular updates, with version 1.0.1 released in November 2023
- Its use of the 3-clause BSD License, ensuring wide accessibility and redistribution rights

## Body
### Development and Version History
SidechainNet was authored by David Ryan Koes. The project is actively maintained, with version 1.0.1 (the current stable version) released on November 2, 2023. Version 1.0.0 was also released on the same date, but version 1.0.1 is marked as preferred. The release notes are available through the GitHub repository.

### Technical Specifications
- **File Format**: Python Pickle serialized data (v4)
- **Based On**: ProteinNet, PyTorch
- **License**: 3-clause BSD License
- **Publication**: Published in the journal Proteins
- **Repository**: Source code available at https://github.com/jonathanking/sidechainnet

### Cited Works
SidechainNet cites several important research papers in the field of protein structure prediction:
- "Structure prediction for CASP8 with all-atom refinement using Rosetta"
- "Improved protein structure prediction using predicted interresidue orientations"
- "Improved protein structure prediction using potentials from deep learning"

### Classification
SidechainNet is classified as both a biological database and open-source software. This dual classification reflects its nature as both a data resource and a computational tool for the scientific community.

### Relationship to Other Projects
SidechainNet is built upon the established ProteinNet dataset and utilizes PyTorch as its underlying machine learning framework. These relationships position it within the broader ecosystem of computational biology tools, leveraging existing infrastructure while adding specialized capabilities for side chain prediction.

## References

1. [Release 1.0.0. 2023](https://github.com/jonathanking/sidechainnet/releases/tag/v1.0.0)
2. [Release 1.0.1. 2023](https://github.com/jonathanking/sidechainnet/releases/tag/v1.0.1)
3. [COCI](https://opencitations.net/index/api/v1/citations/10.1002/PROT.22540)
4. [COCI](https://opencitations.net/index/api/v1/citations/10.1073/PNAS.1914677117)
5. [COCI](https://opencitations.net/index/api/v1/citations/10.1038/S41586-019-1923-7)