What is the best way to organize the query structure?

Question

JavaScript

Task: Get the user's balance, execute some logic and update the balance for this user (or rollback).
The structure is now like this: There is a class with custom logic and there is a class for working with database queries.
Question: Is it normal if I drag a transaction into custom logic like:

@Transaction()
    public async test(userId, userRepository) {
        await userRepository.findOne(userId);
        ... // some logic
        await userRepository.update(userId, ...);
    }

Where some logic is a rather voluminous piece of custom logic.
Or is it better for me to do it in a class with a database and transfer custom logic to it?
After all, I initially had the following project structure: all work with the database is in a separate class, and dragging a transaction into a class that is not intended for this seems to me somehow not right.