I could recommend a step that is frequently performed during annotation of de-novo sequenced transcriptomes (http://trinotate.sourceforge.net/). The idea is to utilize HMM-based domain prediction tools such as Pfam/Hmmer (http://hmmer.janelia.org/software). You can also check out InterProScan (http://www.ebi.ac.uk/Tools/pfa/iprscan5/) which performs domain annotations using a huge variety of methods.