This repository contains code and data created in support of my project, "Local molecular and global connectomic contributions to cross-disorder cortical abnormalities", now published in Nature Communications (and tweeted on twitter).
This project was originally a preprint on bioRxiv, and used to be titled "Molecular and connectomic vulnerability shape cross-disorder cortical abnormalities".
All code was written in Python 3.8.10 and depends on standard Python packages.
This repository can be cloned with git clone https://github.com/netneurolab/hansen_crossdisorder_vulnerability.
The code folder contains the scripts required to conduct the main analyses:
- 00_get_predictors_and_atrophy_maps.py puts together the list of molecular and connectomic predictors used in the main analysis, and puts together ENIGMA-derived case versus control cortical abnormality maps for 13 different disorders/diseases/conditions. I used the enigma toolbox to fetch most of these profiles. This script generates the figures used in Figure 1 of the manuscript.
- 01_regression_and_dominance.py runs the primary analysis of the project, which uses an in-house implemented version of dominance analysis. This script generates the figures used in Figures 2 and 3 of the manuscript.
- 02_network_spreading.py calculates the degree to which disorders demonstrate "network-informed" or "network-spreading" cortical abnormalities, an analysis that was originally developed by Shafiei et al., 2019. This script generates the figures used in Figure 4 of the manuscript.
- 03_disorder_similarity.py compares disorder similarity to other similarity matrices as well as the structural and functional connectomes. This script generates the figures used in Figure 5 of the manuscript.
The data folder contains the data I use to run my analyses. Here I list out the directories and some files of interest:
- enigma contains ENIGMA maps that are (or were) not included in the enigma toolbox. These files come from me scavenging supplementary tables which sometimes come as PDF files and were therefore copy-pasted into something for my scrpit to read. Sorry.
- predictors contains parcellated molecular predictor maps that I use in the main analysis. The volumetric version of many of these maps can be found in the neuromaps toolbox.
- I eventually save my predictor matrices in local_biol_predictors.csv and global_conn_predictors.csv. The names/order of the variables are stored in
{bio/conn}_predictor_names.npy. There are some other versions of the connectivity predictors, as well as MEG-derived temporal predictors, that I use in the supplementary analyses. - I save my region by disorder matrix of cortical abnormalities in enigma_ct.csv and the names/order in disorders.npy.
- Richard Carson and Kelly Smart from the Yale PET Centre shared this PET map, which is of a tracer (UCBJ) that binds to the synaptic vesicle glycoprotein 2A and indexes synapse density.
The results folder contains the saved outputs from the code scripts. Analyses were repeated in Lausanne and HCP SC/FC datasets, hence the duplicated results.