MARS is a gene finding program designed to predict alternatively spliced transcripts de novo. MARS is based on Twinscan (see below for more information).

For further information please contact Paul Flicek: flicek (AT)

MARS is an extension to the Twinscan algorithm for dual-genome conservation-based gene prediction. MARS extends Twinscan version 1.X and participated in the EGASP (the ENCODE Genome Annotation Assessment Project) in 2005.

MARS is a contraction of "Multiple Informants: Alternative Splics" and, described briefly, predicts multi-transcript genes using pair-wise evolutionary conservation models from a number of informant sequences. Details of the algorithm and its performance are described in Flicek & Brent, 2006.

Supplementary material for Flicek & Brent, 2006:

  • Predictions on the ENCODE regions as described in the paper
  • Training set sequences
  • Training set annotation
  • MARS source code

    Mars images copyright NASA. The JPL has many more Mars pictures.
    Last modified: Sun Apr 30 13:56:51 BST 2006