This work is licensed under CC BY-NC-SA 4.0.
Transcriptional and neurochemical signatures of cerebral blood flow alterations in schizophrenia and the clinical high-risk state for psychosis
This repository contains data and code used the project 'Transcriptional and neurochemical signatures of cerebral blood flow alterations in schizophrenia and the clinical high-risk state for psychosis', currently published as a preprint in bioRxiv. The aim of this project is to measure the correspondence between regional cerebral blood flow alterations in the psychosis spectrum and molecular and neurochemical signatures as measured through the Allen Human Brain Atlas and 19 PET tracers.
Code is written in Python.
Folders are described below.
Analysis pipeline. a, 19 Group-level PET maps were obtained from hansenreceptors. b, Group-level case-control rCBF t-stat maps (SSD vs HC, CHR-P vs HC) were calculated in SPM12. c, Gene expression data was obtained from the Allen Human Brain Atlas (AHBA). The AHBA tissue samples of donor brain images were registered to MNI space (ICBM152 nonlinear 2009c symmetric) using Advanced Normalisation Tools. The gene expression data, comprising a total of 17,205 genes, were then pre-processed and log-normalised. d, PET, transcriptomic and case-control rCBF maps were then parcellated into 122 regions of interest (ROI) using standardised brain atlases, with the mean value extracted for each region. Three example distributions (NMDA, CHR-P vs HC, PTPRZ1) are displayed in the figure for illustrative purposes. The resulting averaged distributions within each of the 122 ROIs for each marker (PET map, case-control rCBF map and gene expression map) were z-scored and then subjected to e, Spearman’s correlation to rank all PET maps and genes according to their association with each rCBF marker (SSD vs HC, CHR vs HC, CHR-T vs CHR-NT), and f, dominance analysis to ascertain the unique contribution of each PET map to rCBF phenotypes (SSD vs HC, CHR-P vs HC, CHR-T vs CHR-NT). g, Finally, gene set enrichment analyses were conducted to ascertain specific cell-types involved, using marker genes obtained from a recently published single-cell transcriptomic study from the human brain and filtered by differential stability to retain genes with Pearson’s ρ > 0.529. SSD, schizophrenia spectrum disorders. PET, positron emission tomography. CHR-P, clinical high-risk for psychosis. CHR-T, CHR-P individuals who subsequently transitioned to psychosis. CHR-NT, CHR-P individuals who had not transitioned at follow-up. HC, healthy control. MNI, Montreal Neurological Institute. rCBF, regional cerebral blood flow.
The code folder contains the code used to run the analyses. If you run this code ensure to change the path
-
parcellate_maps_and_generate_nulls will parcellate (e.g. ASL, PET) .nii maps into a given atlas space (e.g. 2mm Schaefer+Xiao 122 ROI atlas). This code uses neuromaps toolbox, PET maps from hansenreceptors, as well as group-level schizophrenia and clinical-high risk for psychosis rCBF t-stat maps we have made available for this project. After parcellating maps, nulls are generated using the burt2020 method, implemented through neuromaps.
-
analysis_correlations will run correlational analysis and generate a heatmap using the comparemaps function, implemented through neuromaps.
-
analysis_dominance will run a dominance analysis adapted from hansenreceptors and implemented through netneurotools.
-
analysis_cooks_distance This will run a Cook's distance analysis of influential ROIs using statsmodels and scipy.
The data folder contains the data used in the analysis including the parcellated data generated from the above code.
- atlas contains MNI152 atlases used for parcellation.
- pet_atlas contains PET maps obtained from hansenreceptors.
- asl contains group-level t-stat maps of rCBF arterial spin labelling case-control comparisions, calculated in SPM12.
The clinical high risk (CHR-P) data was collected as part of two previous studies:
- Resting Hyperperfusion of the Hippocampus, Midbrain, and Basal Ganglia in People at High Risk for Psychosis
- Increased Resting Hippocampal and Basal Ganglia Perfusion in People at Ultra High Risk for Psychosis: Replication in a Second Cohort
The schizophrenia (SSD) data has been presented at these conferences:
- CBF DIFFERENCES BETWEEN HEALTHY AND SCHIZOPHRENIC BRAINS – A FBIRN PHASE 3 MULTISITE STUDY AT 3T USING CBFBIRN DATABASE AND ANALYSIS PIPELINE
- Diagnosis of Schizophrenia using CBF Measures as a Classification Feature – A FBIRN Phase 3 Multisite ASL Study at 3T
Additional maps were created for those CHR-P that subsequently developed psychosis (CHR-T) and those two did not (CHR-NT), as well as maps including age, sex, medication as covariates in the analysis. Participant information can be found in the Participant info table.
The supplementary materials document details scanning parameters and preprocessing.
receptors_parcellated and asl_parcellated .csv files contain asl and pet data that has been parcellated using the 82, 100, and 122 ROI atlases. receptor_names_pet.npy contains a list of pet tracers used in this experiment.
Please refer to the preprint and supplementary materials for a full list of references.
- saved_nulls contains the null maps saved from parcellate_maps_and_generate_nulls.
The figures folder contains figures generated through the above code.
brain render figures were generated through code in python by first extracting each ROI in MNI space using nibabel and then converted to a mesh using nii2mesh. Values for each ROI were input from .csv files finally each combined image was rendered in surfice.
The results folder contains results generated through the above code.
transcriptomic_analysis contains the code and data used for the transcriptomic analysis, implemented through the imaging-transcriptomics package.