Cirq error models

I have two questions about error channels used in Cirq to model noise quantum circuits. The first one is that for the deplorizing error model, one common setting is to set larger error rates for two-qubit gates, such as the CNOT gate, and lower error rates for sigle-qubit gates. How can one achieve that using Cirq?

The second question is after constructing a specific quantum circuit. How to combine two types of error models in the Noise class? Though the offical document have discussed combine multiple errors, but the errors added in the process of constructing the quantum circuits, which are not often the case, especially when one use specific circuit module provided by some pakages.