The ultimate choice stems from your individual interests and not a scientific reason... but having said that some pointer of issues may be handy.
I am assuming that you want to make an inhibitor that messes with the activity and not a ligand that binds somewhere and you then plan to add a moiety to make it a protac to get it degraded or to make it fluorescent to mark it. If not, ignore any talk of activity.
Crystal
- is a crystal structure available?
- does the structure have the cofactor if any?
- is a structure available (itself or homologue) with the native metabolite bound?
- how is the active site? In the case of many cytochrome P450 enzyme the entranceway is really deep and is guarded by a flappy gate, both of which confer a great deal of specificity regardless of the residues in the active site, which are often just water coordinating. Your first enzyme, is that a NADPH/FAD or a cytochrome enzyme?
- What is the mechanism and is there a classic type of inhibitor chemistry? I.e. catalytic triad with a cysteine nucleophile is a nice target.(1)
Methodology
- Wet lab: A problem with computational chemistry/biochemistry is that someone needs to make the protein and physically test your compounds. Are these easy to express? This may sound like not your problem, but a lot of issues can come from this latter stage.
- is there a close homologue you want to avoid targetting? If so, you'll have to do twice the docking (not a big deal and potentially more interesting). (2)
- is the method you plan to use okay with covalently bound ligands? Dopa decarboxylase sounds like a PLP enzyme. If it is, making an inhibitor using classic functional groups that nucleophilically attack the pyridine ring is very straighforward, but not all docking methods are okay with covalent ligands.
- Are any hits likely to follow Lipinski's rule of five? If you have a very charged active site, your hits will be metabolite-like (polar) and not drug-like (horrendously insoluble).
A minor thing is to make sure that nobody else is working on it or you may be left behind. Industry is all cloak and dagger, but the academic SGC has a list of [target enabling packages)[https://www.thesgc.org/tep] and their future ones (nominations).
Physiology
This was most likely in your lit search. But I am double checking.
- You have chosen these enzymes because their unchecked activity is disease-causing, right?
- And they do not have any other roles in other tissues?
- Abolition of activity does what you want?
Other hits
Lastly, but not least. See which of these have many paper about barely active hits: this should mean it is hard. For example, bacterial alanine racemase has been a holy grail for a while, but progress is slow as it has been described as a "ship in a bottle" due to its long and unforgiving entranceway.
Addenda
- A catalytic triad is acid-base-nucleophile: if you have an electrophile with a leaving group (including internal ones, such as epoxydes) you can virtually screen more complex small molecules that resemble your ligand, but with the difference that there is a leaving group, resulting in an irreversible inhibition. Whereas competitive inhibitors require nanomolar affinity to outcompete the natural ligand, an irreversible inhibition can even slightly worse and will work.
- The problem of homologues.
- Paralogues (homologous genes in the same organism): you want to target protein 1, but you do not want to target protein 2. This is a common problem with kinases. You therefore need to find a hit that works better in one than in the other, which with a big of rational inference is easy.
- Homologues (homologous genes in different organism). You want to kill species A (say a fly), but not B (say a human). Same thing.
- Twice the docking. It is more of a challenge, but makes for a better spin on the story. There is an explosion of people doing virtual screens that do not go anyway afterwards, so it is getting harder and harder to publish studies without extensive empirical backing. Hence the need for an interesting angle.
