We developed this facility for bioinformaticians who wish to develop models for predicting function of proteins or their residues. This allows users to create patterns from proteins that can be used directly for developing SVM or ANN based modules. Following are two major modules. For more information click here help

---

Patterns at residue level

This module create pattern at residue level, firstly create overlapping patterns of desired size. Secondly, it assign a pattern positive if central residue of pattern is functional (like DNA interacting), rest of the patterns are assigned negative patterns. Thirdly it calculate features of each patterns and present pattern by numerical values (vector/matrix) readable by standard software packages (e.g., SVM_light, SNNS).

File (annotation at residue level)		Example file: ATP proteins
Window Length: 11 13 15 17 19 21 23 25 27 29	Type of Pattern: Residue Composition Dipeptide Composition Binary profile
Software Package: SVM_light Weka SNNS	Negative pattern: total/all equal to positive

---

Patterns at protein level

This module allow one pattern for one protein, it is useful for develolping model for discriminating two class of proteins (positve vs negative). User need to provide two files one containg positive and another containg negative proteins.

Positive proteins (e.g., ATP binding proteins)
Negative proteins (e.g., Non-ATP binding proteins)
Type of Pattern: Residue Composition Dipeptide Composition	Software Package: SVM_light Weka SNNS