## BioGAIL R Packages

#### R Package for Datamining

At term, the purpose of the "dm" package is to offer an unified framework of datamining tools. In its current realease, it include the following facilities:

- Reading and writing of .arff files (WEKA format) via functions
*read.arff()* and *write.arff()*.
- SMO algorithm to solve Support Vector Machine (SVM) problems, with user defined kernels. This is done by function
*smoSolver()* which take as input the Gram matrix of training instances and solve the maximal margin hyperplane problem. One can also used *smo()* and *predict.smo()* for extended functionalities.
- Kernel functions:
*polynomial.kernel()*
- Attribut ranking by Support Vector Machine Ranking with linear machines or LogRatio machines (see function
*svmRank()*).
- Recursive Feature Elimination Ranking via
*rfeRank()*. When used in combination with *svmRank()*, one can build an svmRfe attribut ranking.
- Splitting of a dataset into stratified folds (see
*stratifiedSplit()*)

Download: dm_3.3.tar.gz

Contact: julien.prados@cui.unige.ch

#### R Package for Mass Spectrometry

The "ms" package contains functions to process mass spectrometry data. It contains:

- Sliding windows filtering and smoothing functions for 1D signals:
*avgfilt()*, *closefilt()*, *percfilt()*, *sdfilt()*
- Peak detection functions for 1D, 2D, nD signals:
*valleys()*, *valleys2D()*, *valleysGraph()*
- A method to build delaunay triangulation graph of a set of 2D points:
*delaunay()*
- A linear model fitting method on 1D peaks
*lmPeakFit()*
- A peak alignement/clustering method
*mzcl()*
- A function for hierachical clustering comparison
*hclustdiff()*
- An utility function to compute areas of triangles
*triangle.area()*

Contact: julien.prados@cui.unige.ch