Skip to content

Home

Jump to bottom
LiangfangLi edited this page Nov 21, 2021 · 2 revisions

Welcome to the ISVGeneExpression wiki!

Abstract

It has been revealed that intersubject variability (ISV) in intrinsic functional connectivity (FC) is associated with a wide variety of cognitive and behavioral performances. However, the underlying organizational principle of ISV in FC and its related gene transcriptional profiles remain unclear. Using resting-state fMRI data from the Human Connectome Project (299 adult participants) and microarray gene expression data from the Allen Human Brain Atlas, we conducted a transcription-neuroimaging association study to investigate the spatial configurations of ISV in intrinsic FC and their associations with spatial gene transcriptional profiles. We found that the multimodal association cortices showed the greatest ISV in FC, while the unimodal cortices and subcortical areas showed the least ISV. Importantly, partial least squares regression analysis revealed that the transcriptional profiles of genes associated with human accelerated regions (HARs) could explain 31.29% of the variation in the spatial distribution of ISV in FC. The top-related genes in the transcriptional profiles were enriched for the development of the central nervous system, neurogenesis and the cellular components of synapse. Moreover, we observed that the effect of gene expression profile on the heterogeneous distribution of ISV in FC was significantly mediated by the cerebral blood flow configuration. These findings highlighted the spatial arrangement of ISV in FC and their coupling with variations in transcriptional profiles and cerebral blood flow supply.

Software and Toolbox Requirement Statement

Software:

  1. The Matlab scripts have been tested on Matlab 2018b. It has not been tested for any other Matlab versions.

  2. We generated spatial-autocorrelation-preserving surrogate maps using the BrainSMASH software (Brain Surrogate Maps with Autocorrelated Spatial Heterogeneity, Burt et al., 2020). You can visit https://brainsmash.readthedocs.io/en/latest/ to get more information.

Toolbox:

  1. The partial least squares (PLS) regression analyses in this project were performed by using the code shared by Whitaker et al. (2016). You can visit https://github.com/KirstieJane/NSPN_WhitakerVertes_PNAS2016 for more information.

  2. The gene-category enrichment analysis (GCEA) in this project was performed by using a recently released Matlab toolbox (Fulcher et al., 2021). You can visit https://github.com/benfulcher/GeneCategoryEnrichmentAnalysis for more information.

  3. fMRI data were processed by using the Data Processing Assistant for Resting-State fMRI (DPARSF) (Yan and Zang, 2010; Yan et al., 2016; http://rfmri.org/DPARSF) and Statistical Parametric Mapping software (SPM12; http://fil.ion.ucl.ac.uk/spm).

  4. We used the toolbox “NIfTI_20140122” to load, view and save imaging data. You can download it from https://github.com/Alxemade/NIfTI_20140122.

Running Analysis

The following is a detailed explanation of the codes including measures calculation, statistical analysis, and figure generation. Run the following scripts in sequence:

Step_1st_ISV_Calculation

  1. Step_1st_calculate_ROI_level_ISV.m:

    calculate intersubject variability in functional connectivity (ROI-level).

  2. Step_2nd_calculate_WithinModule_level_ISV.m:

    calculate intersubject variability in functional connectivity (within-module level).

Step_2nd_GeneExpression_Calculation

  1. Step_1st_average_probes_with_SameGeneSymbol.m:

    average expression values of multiple probes corresponding to the same gene symbol to generate gene expression levels for each sample.

  2. Step_2nd_normalize_expression_by_SampleMean.m:

    normalize gene expression values of all genes on each sample by dividing the average gene expression value of the sample.

  3. Step_3rd_select_left_hemesphere_sample.m:

    select samples located in the left hemisphere according to the sample annotation in SampleAnnot.csv.

  4. Step_4th_find_nearst_voxel_forSample.m:

    calculate the minimal Euclidean distance between the reported MNI coordinates of samples and that of all gray matter voxels, find the closest voxel for each sample.

  5. Step_5th_average_sample_with_SameROILabel.m:

    average the expression data of the samples mapped to the same particular ROI.

  6. Step_6th_normalize_expression_across_ROIs.m:

    within each donor, gene expressions were normalized to Z scores across all cortical regions per gene.

  7. Step_7th_average_donors_with_SameROILabel.m:

    average normalized gene expression profiles across 6 donors to obtain a group-level gene expression matrix.

  8. Step_8th_extract_HARBRAIN_BRAIN_expression.m:

    extract gene expression matrix of HAR-BRAIN genes and BRAIN genes (these two gene lists were obtained from Wei et al., 2019).

Step_3rd_PLS_regression

The partial least squares (PLS) regression analyses in this project were performed by using the code shared by Whitaker et al. (2016). You can visit https://github.com/KirstieJane/NSPN_WhitakerVertes_PNAS2016 for more information.

  1. Step_1st_prepare_variable.m:

    prepare predictor variable and response variable for PLS regression.

Step_4th_Specificity_Analysis

  1. Step_1st_SpecificityAnalysis_BRAINgene.m:

    validate the specificity of HAR-BRAIN genes based on the BRAIN gene.

  2. Step_2nd_SpecificityAnalysis_AHBAgene.m:

    validate the specificity of HAR-BRAIN genes based on the AHBA genes.

Step_5th_GeneCategoryEnrichmentAnalysis

We performed gene-category enrichment analysis (GCEA) by using a recently released Matlab toolbox (Fulcher et al., 2021). You can visit https://github.com/benfulcher/GeneCategoryEnrichmentAnalysis for more information.

Step_6th_CerebralBloodFlow

  1. Step_1st_generate_CBFmap.m:

    extract CBF value from the cerebral blood flow map shared by Satterthwaite et al. (2014) to obtain an ROI-level CBF map based on AAL-625 atlas.

  2. Step_2nd_combine_pls_isv_cbf_forMediation.m:

    combine PLS score, CBF and ISV score to conduct mediation analysis.

Step_7th_Mediation

We performed bootstrapped mediation analysis using the PROCESS macro in SPSS (Hayes, 2017).

Step_8th_Generate_SurrogateMap

We generated spatial-autocorrelation-preserving surrogate maps using the BrainSMASH software (Brain Surrogate Maps with Autocorrelated Spatial Heterogeneity, Burt et al., 2020). You can visit https://brainsmash.readthedocs.io/en/latest/ to get more information.

  1. Step_1st_generate_BRAINSMASH_input_files.m:

    generate input files for using BRAINSMASH.

  2. Step_2nd_generate_SurrogateMap_BRAINSMASH.py:

    Python code to generate surrogate brain maps by using the BRAINSMASH software.

Step_9th_Visualization

  1. Step_1st_plot_histogram_forSpecificityAnalyses.m:

    plot histogram for the specificity analysis based on BRAIN genes and AHBA genes.

  2. Step_2nd_plot_ScatterForCorrelationAnalyses.R:

    R code for plotting the scatter plot.

Clone this wiki locally