October 10, 2004

Turning numbers into real biology: tissue specificity of alternative splicing

With large number of alternative splicing events detected from human and other genomes, one immediate question is: are these alternative splicing events regulated?

Here is a list of papers on tissue-specificity of alternative splicing:

Genome-wide detection of tissue-specific alternative splicing in the human transcriptome. Xu Q, Modrek B, Lee C. Nucleic Acids Res. 2002 Sep 1;30(17):3754-66.

Variation in alternative splicing across human tissues.Yeo G, Holste D, Kreiman G, Burge CB. Genome Biol. 2004;5(10):R74.

Strengths and weaknesses of EST-based prediction of tissue-specific alternative splicing. Gupta S, Zink D, Korn B, Vingron M, Haas SA.
BMC Genomics. 2004 Sep 28;5(1):72.

The basic idea for computational identification of tissue-specific alternative splicing, is to count number of EST evidences for each splice variants from different tissues, and look for evidence of tissue-specificity.
For example, Xu et.al. used the following strategy: for each pairs of alternative splices(S1,S2), and a given tissue(T), they counted the number of ESTs in tissue T which support either S1 or S2(N1,N2), as well as the number of ESTs which support either S1 or S2 in all other tissues(N1',N2'). Then they tested whether the frequency of splice S1 in tissue T was significantly greater than the frequency of splice S2 in tissue T, using bayesian inference.

Here is an Excel spreadsheet I made for a bioinformatics class at UCLA, which demonstrates the computation of Xu et.al. . Click to download.

Posted by yxing at October 10, 2004 02:20 AM
Comments

This turn is really exciting,for here we can see some practical uses of bioinformatics analysis of AS which can be apt to the bench workers thoughts and gives a new perspective for us.
Question: what is the current focusing of AS except its tissue specificity,any other specificity interesting ?

Posted by: yummy at October 11, 2004 10:47 AM

Alternative splicing can also be tightly regulated by developmental states. It can also have strong association with diseases ( e.g. cancer). All these are interesting subjects for studies.

Posted by: Yi Xing at October 11, 2004 04:17 PM

Nice response. And since you claim that AS is related to such as development,dieases, etc, then can computation be useful for these work? And how?

joke: only statistics?

Posted by: lylover at October 12, 2004 04:03 PM

sure. if someone classifies the EST libraries based on developmental states, he can do the same kind of analysis for detecting tissue-specificity, as long as there is enough EST counts for libraries with development-state information.

And experimental work can follow computational discoveries.

Posted by: Yi Xing at October 12, 2004 04:18 PM