Even without introducing new amino acids (which sounds very tricky indeed, as the article notes, this could really mess up normal biological structure and function due to codon usage issues, i.e. if you tried to devote some of the 61 available codons (64 - 3 stop codons) to unique amino acids, ribosomes might start plugging the weird ones into normal proteins with catastrophic results. Making a mirror-image ribosome seems like a lot of work... and with some issues, like escaping into the wild might be a problem, i.e. if a mirror-image microbe started replicating, what would eat it? Would it be the worst invasive species imaginable?<p>There are more standard approaches that are interesting, like extending existing biochemical synthesis pathways via novel proteins, such as microbes capable of standard chemical conversions of natural products, such as salicylic acid + acetin anhydride to aspirin and similar transformations of more interesting compounds. Usually these are small-molecule additions to large natural products, so you have to engineer the metabolism to make the small molecule at the same rate as the large molecule and then build a protein capable of the final synthetic step (using only the standard set of amino acids).