Tnasas

	-
	Tnasas

Tnasas: a class prediction and gene selection tool

Tnasas is a web tool for building class prediction models (but, as the name says, this is not a substitute for a statistician). We include five class prediction models/algorithms and three gene selection methods/algorithms. We pay particular attention to obtaining honest estimates of the prediction error rate and assessing instability of solutions. For further details see the help.

To use this web tool you need to provide two files, one with the gene expression data and one with the class labels, and press "Submit".

Input files

Expression data file: (?)
Class file: (?)

Class prediction algorithm and gene selection method

	Classification algorithm	Gene selection method
	DLDA (?)	F ratio (?)
	KNN (?)	Wilcoxon statistic (?)
	Random forest (?)	Random forest (?)
	SVM (?)
	Shrunken centroids (PAM) (?)

Push the "Submit" to start execution.

Help

Tnasas has been developed by Juan M. Vaquerizas, currently at the EBI, and Ramon Diaz-Uriarte, at IIB and UAM. It is maintained (kept alive) by Ramon Diaz-Uriarte.

Send comments to webmaster

Last modified: October 2014

UPDATE and disk image availability: Minor updates fixing/updating some links, improving some of the logic, using R-4.5.1, and allowing to use https, on September and October 2025. I have a QCOW2 disk image containing pomelo2, signs2, genesrf, and tnasas, along with a libvirt XML configuration file. This would let you run the environment locally using KVM/QEMU and modify the setup as needed. Let me know if you'd like a copy. (Note, though, that most of the code is from over 15 years ago, and it is still using Python 2.7. The code would benefit from a major clean up, so this was just a minimal, hackish intervention to allow reviving the apps on a virtual machine).