A more serious post, at last!

My boss just pointed me out an article that was published in last April's issue of Bioinformatics (See References). I thought I would write a small post about it. Its authors describe the new bioinformatic tool that they developed to predict phosphorylation sites of yeast proteins.

The tool's name is NetPhosYeast. it was created using datasets of yeast proteins known phosphorylation sites to train an artificial neural network, which is now available for yeast researchers. Just follow the link. The authors claim that their tool has a 90% specificity (true hits to total hits) and a 84% sensitivity (hits to total number of true sites), which seems amazing to me.

I just got my favorite proteins through the algorithm, and got encouraging answers... I'm not sure yet of their validity, but it seems I'm on the right track which is encouraging!

References

Bioinformatics, Vol. 23, No. 7. (1 April 2007), pp. 895-897.