For example, a kind of enzyme is characterized by the N-terminal Serine/Threonine kinase. And my question is how to find the kinase activity site, how to make the kinase dead with one site mutation without any reference.

There are several databases that can assist you in that, though of course without actual experimental verification they all will just be predictions based on sequence similarities. Uniprot is a general database that aggregates a lot of this information. If you want to implement something yourself you can look into protein family/domain databases such as pfam, prosite, etc.

But again, usually you would knock out the whole domain, or at least a sizeable chunk of it. Figuring out the required minimum mutation to deactivate a given activity is quite a bit trickier.

There are several databases that can assist you in that, though of course without actual experimental verification they all will just be predictions based on sequence similarities. Uniprot is a general database that aggregates a lot of this information. If you want to implement something yourself you can look into protein family/domain databases such as pfam, prosite, etc.

But again, usually you would knock out the whole domain, or at least a sizeable chunk of it. Figuring out the required minimum mutation to deactivate a given activity is quite a bit trickier.

Thanks for your reply