I can think of at least two strategies.
First, split your 753 proteins into single files, do BLASTp searches for each of them against your ORF database using the E-value threshold of
0.00001, and save the output of each search into individual files. BLASTp search files with no matches have the size of about 2500 characters - you can find the exact range of numbers easily by finding out the smallest 5-10 files after the search. After that there will be a jump in file sizes of at least 500, which are the files that contain matches. NOTE: if your ORF database is smaller than 753 proteins, you may want to do a reverse strategy: split your database into individual proteins, and BLASTp against the Bacmet ORFs.
Run jackhmmer, HHpred or some other automated search strategy on your Bacmet proteins and build HMM files for each of them. Concatenate all HMMs into a single file, and annotate your ORFs against this database using Prokka.