At last our paper on predicting PDZ binding has been published. In this work we built a custom machine learning model to identify peptides that are likely ligands of a given PDZ domain.
PDZs are a very common structural protein units that are involved in cell signalling pathways. The binding occurs in a cleft between an alpha helix and beta sheet, typically by turning the ligand into an additional strand at the end of the sheet.
In this work we showed how the number of false positive predictions could be reduced by making a consensus predictor that combines a number of individual machine learning components. They are joined by a voting mechanism, and then made turned into a rational number using probability distributions over the residues at each of the binding positions.
Unfortunately the process is too complicated to automate at this stage.
You can find it here:Reduced False Positives in PDZ Binding Prediction using Sequence and Structural Descriptors
PDZs are a very common structural protein units that are involved in cell signalling pathways. The binding occurs in a cleft between an alpha helix and beta sheet, typically by turning the ligand into an additional strand at the end of the sheet.
In this work we showed how the number of false positive predictions could be reduced by making a consensus predictor that combines a number of individual machine learning components. They are joined by a voting mechanism, and then made turned into a rational number using probability distributions over the residues at each of the binding positions.
Unfortunately the process is too complicated to automate at this stage.
You can find it here:Reduced False Positives in PDZ Binding Prediction using Sequence and Structural Descriptors
No comments:
Post a Comment