A Knowledge Base for Predicting Protein Localization Sites in Eukaryotic Cells

A Knowledge Base for Predicting Protein Localization Sites in Eukaryotic Cells

1992 | KENTA NAKAI AND MINORU KANEHISA
The article by Kenta Nakai and Minoru Kanehisa from Kyoto University's Institute for Chemical Research presents an expert system designed to predict protein localization sites in eukaryotic cells based on amino acid sequences and source origin. The system is built using a knowledge base of if-then rules, which organizes various experimental and computational observations. The authors collected data for 401 eukaryotic proteins with known localization sites and divided them into training and testing datasets. The system successfully predicted 66% of the training data and 59% of the testing data. The article discusses the challenges and limitations of the approach, including the need for more specific knowledge about certain sorting pathways and the difficulty of assigning a single localization site to each protein. The system's performance is evaluated against widely used protein secondary structure prediction, showing better accuracy. The authors conclude that the system is flexible and can incorporate diverse types of sorting signals, but improvements are needed in knowledge acquisition and maintenance.The article by Kenta Nakai and Minoru Kanehisa from Kyoto University's Institute for Chemical Research presents an expert system designed to predict protein localization sites in eukaryotic cells based on amino acid sequences and source origin. The system is built using a knowledge base of if-then rules, which organizes various experimental and computational observations. The authors collected data for 401 eukaryotic proteins with known localization sites and divided them into training and testing datasets. The system successfully predicted 66% of the training data and 59% of the testing data. The article discusses the challenges and limitations of the approach, including the need for more specific knowledge about certain sorting pathways and the difficulty of assigning a single localization site to each protein. The system's performance is evaluated against widely used protein secondary structure prediction, showing better accuracy. The authors conclude that the system is flexible and can incorporate diverse types of sorting signals, but improvements are needed in knowledge acquisition and maintenance.
Reach us at info@study.space
Understanding A knowledge base for predicting protein localization sites in eukaryotic cells