it is impossible to predict function insilico, but expectation is very weak for the protein to maintain the proper function. ALl depends on the splice site.

you can try, or do a combination of complementary in silico tools to evaluate the mutations. just as Beston Nore said, it's difficult.

I having the same problem......

http://www.umd.be/HSF/#

but one can not defend this mutation without expression studies.