[anigbrowl]

Read good books, but do so away from the computer; get used to running small-medium chunks of code in your head. I don't want to recommend any in particular because different styles of thinking and writing appeal to different people, but I suggest avoid the ones that emphasize speed or simplicity in the title.

Also, read lots of scientific papers. You'll learn all kinds of interesting stuff, but you'll also need to think about how you would implement their process if a scientist asked you for programming help. Nowadays many papers also come with access to their methods and/or a Github repository.

Because it's so common, you should learn some basic statistics in order to be able to better read things in context, even if you decide to use external tools or libraries to do statistics. I found 'Data Analysis with Open Source Tools' by Philipp Janert (O'Reilly) a great resource: it's language-agnostic, grounded in real-world practices, and assumes you want to solve problems using statistics (and other mathematical analysis tools) rather than treating the math as an end in itself. He includes much valuable context on how and why statistical concepts have evolved, and when you should stick with convention or experiment playfully. Most importantly, he enjoys math and does a good job of communicating his enthusiasm to the reader.