The standard reference for thematic analysis is Braun & Clarke (2006), but my personal opinion is that they are rather vague about how to develop the actual codes.

If you do not have a prior coding system for a deductive or top-down approach to coding, then you need to develop your own inductive or bottom-up coding system. Developing your own codes can be an intimidating process in the beginning, but it becomes much easier as you continue to work with the data. In particular, a common approach is to begin with a high number of "small" codes and then group similar codes into "larger" categories.

Here are links to two short articles on coding. These both use ATLAS.ti as their preferred software, but the process would be basically the same with any of the major programs:

http://atlasti.com/2017/03/16/make-best-codes-atlas-ti/

http://atlasti.com/2017/07/01/gt/

If you like those, the same author has a 60-minute video tutorial on coding:

https://www.youtube.com/watch?v=EvcQLhEZP-A

I have used a simple open coding method in my qualitative papers if it helps, detailed in the method section.

Article Sources of Self-Efficacy in Springboard and Highboard Diving...

Article Barriers and facilitators to overcoming mental block in spri...

We were confronted with the same question in our research project and we finally adopted a method of coding combining inductive and deductive codes. We wrote about it in this paper. I hope it will be helpful for you.

Chapter A Method Combining Deductive and Inductive Principles to Def...