People certainly have made their own Y2H libraries, but it is very challenging.  You have to make cDNA from the relevant cell type, perform second strand synthesis, incorporate linkers on both ends, then ligate the DNA into the Y2H vector and transform E. coli.  Given that different transcripts vary tremendously in abundance and that many of the inserts will be out of frame, you need to get millions of colonies out of this to be confident that the library incorporates all the proteins you will want to screen.  You also have to get the plasmids out of E. coli and into yeast without losing complexity.  If a library exists from the source you are interested in, it may be more cost effective to buy it.  I would recommend the "mate & plate" approach in which you buy yeast already transformed with the library and mate them to yeast transformed with your construct.

Thank you for the answer. My problem is that the organism I'm working doesn't have an exsiting library. But your answer made me rehink the wholw idea.