The rest of this book is organized as reference material — datasets in one place, tools in another, methods in a third. That works well once you know what you are looking for. It works badly on your first week.
This section is the other view: the same documents, arranged in the order you will actually need them. Each step is short and mostly links. The depth lives in the reference pages; this is just the thread that runs through them.
The path¶
Find data — which open datasets exist, and how to get access
Get compute — an account on a cluster, and somewhere to run code
Organize your data — BIDS, and why converting early saves weeks
Preprocess — QC and the standard pipelines
Analyze — the methods scholars in this program actually use
Share your work — git, environments, containers
A word on order¶
The sequence matters more than it looks. The two most common ways a project stalls in this field are both ordering problems:
Preprocessing before organizing. Almost every pipeline worth using expects BIDS as input. Converting first is a day; retrofitting it after you have written analysis scripts against a bespoke layout is a fortnight.
Requesting data before requesting compute. Access applications and cluster allocations both take real calendar time and are independent of each other. Start both early, in parallel.
You do not have to start at step 1¶
If you already have data and a cluster account, skip to organize your data. If you are only here to find out which method other scholars used for a problem like yours, go straight to analyze — or browse the research topics and find the scholar working closest to your question.