Interaction of Manual and Automated Testing in a Project

A discussion on how to build interaction between manual and automated testing in a project. Who will write cases? How to plan the work of the testing team? Should there be a single team or should automation testers be allocated to a separate team?