What is the best folder organization for a large data analysis project with multilayered subprojects?

My project involves several subprojects. They should be completed in a sequential manner. And each subproject involves different steps of data wrangling and analysis. Each step would produce intermediate data, which are needed for the next step. I have two folder tree design. I don't know which one is better. What would be potential pros and cons for each design?

The first tree design.

subproject 1

step 1 (code, data, figure, ...)

step 2 (code, data, figure, ...) uses data from step 1 as input and output new data

step N (code, data, figure, ...)

subproject 2, ..., subproject N

each has multiple steps and similar sub-folder-tree

The second tree design.

code (subproject 1 (step 1, step 2, ...), subproject 2 (step 1, step 2, ...) , ..., subproject N)

data (subproject 1 (step 1, step 2, ...), subproject 2 (step 1, step 2, ...), ..., subproject N)

figure (similar sub-tree pattern)

[Simon]

I have installed the tree class and am just about to play with it.