Predicting Intestinal Permeability of Peptides Using Neural Networks
Author Information
Author(s): Jung Eunkyoung, Kim Junhyoung, Kim Minkyoung, Jung Dong Hyun, Rhee Hokyoung, Shin Jae-Min, Choi Kihang, Kang Sang-Kee, Kim Min-Kook, Yun Cheol-Heui, Choi Yun-Jaie, Choi Seung-Hoon
Primary Institution: Insilicotech Co. Ltd.
Hypothesis
Can artificial neural networks effectively predict the intestinal permeability of oligopeptides based on their sequence information?
Conclusion
The study successfully developed models that can predict the intestinal permeabilities of oligopeptides based on their sequences.
Supporting Evidence
- The models were validated using statistical indicators like sensitivity and specificity.
- Training and test set statistics indicated high quality models capable of distinguishing permeable from impermeable peptides.
- Models showed better predictive power with simpler architectures compared to complex ones.
Takeaway
Scientists created computer models to guess how well certain tiny proteins can pass through the gut, which helps in making better medicines.
Methodology
The study used artificial neural networks trained on peptide sequences to predict intestinal permeability.
Potential Biases
The reliance on random sequences as negative controls may introduce bias in the predictions.
Limitations
The model may not accurately predict permeability for all peptide sequences, especially those not represented in the training data.
Statistical Information
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website