I think there is no general answer to your question.
Which type of ID you will use depends on how practical it will be for your downstream processing. This depends on what other source of information you have, how much conversion work you will need to do in order to match your data to these information if you use one or the other.
I think the choice may emerge by trial and error, as you realize that a given choice is actually not convenient in your particular use case. Try to set up easy to use conversion procedures that you can easily insert in some steps of your workflow, so that you can easily switch from one choice to the other.
Concerning the tools to do the conversion, see the following question and its answers: Converting gene names from one public database format to another