GAMBA is a web-based application (www.dutchconnectomelab.nl/GAMBA/) that can be used to test whether the gene expression profile(s) of the input gene(s) and neuroimaging-derived brain features show overlapped spatial patterns. Different statistical null models are availale to examine both gene specificity and spatial specificity.
For details, please see:
Wei Y. et al., Statistical testing and annotation of gene transcriptomic-neuroimaging associations, in preparation.
Wei, Y. et al. (2019), Genetic mapping and evolutionary analysis of human-expanded cognitive networks. Nat Commun. https://doi.org/10.1038/s41467-019-12764-8
GAMBA is a web-application with a front-end interface showing pre-processed results on associations between gene transcription and imaging-derived brain features. Processing steps and statistical analyses are included in /processing
Briefly, given an input gene expression data matrix (region by gene) and phenotypic data matrix (region by phenotype), linear regression is performed to first examine the spatial overlap between gene expression and the phenotypic profile. Then, the null-random, null-brain, and null_coexpression models are generated by randomly sampling genes from all genes, all brain-expressed genes, and genes with similar coexpression level, respectively. The null-spin model is generated based on randomized gene expression matrices according to the spinned brain parcellation.
To get started, please see /processing/README.txt for details. Please note that it may be computational costly to finish all processing. Processed data used for the three examples (see below) are included.
We use three simple examples that show analyses commonly performed in literature to illustrate the usage of different statistical null models. Examples include human-supragranular-enriched (HSE) genes, APOE gene, and risk genes of autism spectrum disorder (ASD). To get started, please see /examples/README.txt for details.
- Example of associations between the spatial pattern of HSE gene expression and the connectome metrics:
/examples/scripts_example1_HSE.m
- Example of associations between the spatial pattern of APOE gene expression and the pattern of brain atrophy in diseases:
/examples/scripts_example2_apoe.m
- Example of associations between the spatial expression pattern of ASD risk genes and the pattern of functional changes in diseases:
/examples/scripts_example3_ASD.m
We simulate and analyze the outcome of different statistical evaluation approaches for a wide range of real brain phenotypes and artificial gradients. To get started, please see /silumation/README.txt for details.
GAMBA is developed using Vue.js. Source code can be found at https://github.com/yongbin-wei/webgamba.git.