I think you can predict functional profile for a specific genera using PICRUSt, for that you just provide one OTU as an input which belongs to desired genera. Another easy way is that you can open "ko_18may2012_precalculated.tab.gz (large file)"(http://kronos.pharmacology.dal.ca/public_files/picrust/picrust_precalculated_v1.1.3/18may2012/ko_18may2012_precalculated.tab.gz) file and extract KEGG profile for a specific OTU (genera) of your choice.
Reference database used in PICRUSt was based on OTU ids as per 18may2012 version of Greengenes database. If you want to predict functions for a specific genera i.e. Bifidobacterium, you need to get OTU id first. Using that OTU id make a input file and predict functions using "predict_metagenomes.py" script available in PICRUSt.
I tried this for Bifidobacterium (OTU id: 49417 for "k__Bacteria; p__Actinobacteria; c__Actinobacteria; o__Bifidobacteriales; f__Bifidobacteriaceae; g__Bifidobacterium; s__") and it worked well.
I am attaching input and output files for your reference. I hope it would be helpful for you.