Skip to content

davidanastasiu/antibiofilm

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Identification of distinct characterisitcs of antibiofilm peptides and prospection of diverse sources for efficacious sequences

This respository contains data and code associated with the paper found here

Authors

Bipasa Bose, Taylor Downey, Anand K. Ramasubramanian, and David C. Anastasiu

Summary

In this work, we developed machine learning models to identify the distinguishing characteristics of known antibiofilm peptides, and to mine peptide databases from diverse habitats to classify new peptides with potential antibiofilm activities. Additionally, we used the reported minimum inhibitory/eradication concentration (MBIC/MBEC) of the antibiofilm peptides to create a regression model on top of the classification model to predict the effectiveness of new antibiofilm peptides. We used a positive dataset containing 242 antibiofilm peptides, and a negative dataset which, unlike previous datasets, contains peptides that are likely to promote biofilm formation. We utilized our classification-regression pipeline to evaluate 135,015 peptides from diverse sources and identified antibiofilm peptide candidates that are efficacious against preformed biofilms at micromolar concentration.

Data

MBIC training data can be found here
MBEC training data can be found here
Test peptides can be found here

MBIC/MBEC scripts

Two independent models were developed in order to predict the minimum inhibitory/eradication concentration (MBIC/MBEC) of unknown antibiofilm peptides. The MBIC model uses an SVM to split the peptides first, applying an SVR to predicted peptides that are less than 64uM. The MBEC model directly trains an SVR on the MBEC peptides. Please see the paper for more details. Both the MBIC and MBEC folders contain source code to train the models which sweep through different hyperparameters and perform forward selection. Both folders contain the optimized features found during forward selection (.json file) and a script that uses the found hyperparameters/features to make predictions about unknown peptides. The MBIC folder contains an additional script that combines both the tuned SVM and SVR which are then used to perform cross-validation and return an average RMSE of the model.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published