Hi, Feyzollah, 

in papers comparing the performance of splice site prediction tools, the Maximum Entropy model usually occupies the forefront positions. Next, there use to be HSF and SSF. There is a disadvantage of MaxEnt tool - with most online tools you have to designate the exon-intron boundary you wish to predict with the MaxEnt. Fortunately, this can be overcome by doing MaxEnt predictions on the HSF website (the new version allow this: http://www.umd.be/HSF3/HSF.html).

Below, I add several links - to recommend some of the papers that describe the performance of splice sites predictors.

Best regards, 

Lucie

http://www.ncbi.nlm.nih.gov/pubmed/22505045

http://www.ncbi.nlm.nih.gov/pubmed/18449911

http://www.ncbi.nlm.nih.gov/pubmed/17576681

http://www.ncbi.nlm.nih.gov/pubmed/16963498

Dear Feyzollah,

we have a splice site prediction tool online for free public use:

https://varSEAK.bio

For Donor / Acceptor splice sites NetGene2 Server - DTU was pretty accurate. Bioinformaticshttps://services.healthtech.dtu.dk/services/NetGene2-2.42/

http://spliceport.cbcb.umd.edu/

https://www.fruitfly.org/seq_tools/splice.html

https://services.healthtech.dtu.dk/services/NetGene2-2.42/