autism-research-centre / autism-gradients Goto Github PK

Now its all in one long script, section need to be cleanly parcellated to different notebooks so its easier to update and debug at different stages by different people.

Also save the output intermediately so that the next notebook can actually start from there.

Owen would you be able to give some guidance on how figures should be embedded in the notebook markdown files?

Upload raw time series data for the relevant subject to github so people can run everything self-contained from the start.

Also include the phenotype file.

From @rb643 on February 28, 2017 11:31

Run neurosynth meta-analysis for binnen gradients (see Fig 4. http://www.pnas.org/content/113/44/12574 )

Copied from original issue: rb643#6

From @rb643 on December 1, 2016 15:48

figure out how to compare gradient slopes

Copied from original issue: rb643#3

From @rb643 on March 2, 2017 10:46

Possibly try and create some plots from the descriptives in the excel file:

http://blog.bharatbhole.com/creating-boxplots-with-matplotlib/

Copied from original issue: rb643#8

Need to mention installing the matlab python engine and registering the matlab_kernel with jupyter

Work out reverse inference of neurosynth keywords based on the original paper

Some subjects were excluded in the original paper due to various reason, these need to be excluded from the present analysis as well, right from the start.

For some strange reason the subsetting at the beginning of 5c return an array of NaN's even though the csv definitely contains normal integers...

From @rb643 on February 28, 2017 11:28

Create a GoF measure to compare the obtained gradients for every individual to known networks

Copied from original issue: rb643#5

Binning the principal gradient into 10 percentile block and create nifti's for each. Then compare each bin against the same neurosynth keywords as were used in the original paper.