Explicit alternative splicing prediction