Behavioral Variance Explained by Neuroimaging and Methylation Data via Pattern-Based Regression
Poster No:
691
Submission Type:
Late-Breaking Abstract Submission
Authors:
Ao Li1
Institutions:
1Fudan unniversity, Shanghai, China
First Author:
Introduction:
Understanding the relationship between neuroimaging and human behavior is a fundamental goal of neuroscience. Epigenetic modifications, particularly DNA methylation (DNAm), dynamically regulate gene expression (Moore et al., 2013), affecting both brain and behavior. Traditional neuroimaging analyses, such as voxel- and cluster-level association studies (Eklund et al., 2016; Gong et al., 2018), often focus on localized brain regions, potentially overlooking distributed neural influences. Similarly, complex behaviors likely emerge from subtle, widespread effects across the brain (Zhao et al., 2021), much like the cumulative impact of distributed epigenetic modifications. Thus, a holistic framework for simultaneously estimating brain-associated variance (BAV) and DNAm-associated variance (DmAV) is needed to integrate whole-brain mapping with epigenetic regulation.
Inspired by LD score regression (Bulik-Sullivan et al., 2015), we introduced a voxel-based brain-wide pattern regression and a site-based methylation-wide pattern regression to provide an unbiased estimation of BAV and DmAV, revealing shared neural and genetic bases across mental disorders and cognitive behaviors.
Inspired by LD score regression (Bulik-Sullivan et al., 2015), we introduced a voxel-based brain-wide pattern regression and a site-based methylation-wide pattern regression to provide an unbiased estimation of BAV and DmAV, revealing shared neural and genetic bases across mental disorders and cognitive behaviors.
Methods:
We proposed the voxel dependence index (VDI) for brain-wide patterns and the DNA methylation dependence index (DmDI) for methylation-wide patterns, defining their scores similarly to LD scores. The imbalance of sparsity between blocks in the DNAm network affects the estimation of overall heritability. We used DmDI scores to identify the most influential sites, sparsified the network accordingly, and obtained the DmDI sparse pattern for regression.
To evaluate the performance of VDI pattern and DmDI sparse pattern regression models, we conducted 1000 signal simulations across different signal-to-noise ratios to assess errors between observed and expected values. Two linear regression steps estimated brain-associated covariance (BAC) and DNAm-associated covariance (DmAC), leading to neuroimaging and epigenetic correlation calculations.
We applied our models to neuroimaging, genetic, and behavioral data from the IMAGEN project, which integrates brain imaging and genetics to study biological, psychological, and environmental influences on mental health. Based on prior studies (Jia et al., 2020), we selected 19 behavioral measures, including ADHD, depression, alcohol consumption, and smoking, to compute their BAVs and DmAVs (Figure1).
To evaluate the performance of VDI pattern and DmDI sparse pattern regression models, we conducted 1000 signal simulations across different signal-to-noise ratios to assess errors between observed and expected values. Two linear regression steps estimated brain-associated covariance (BAC) and DNAm-associated covariance (DmAC), leading to neuroimaging and epigenetic correlation calculations.
We applied our models to neuroimaging, genetic, and behavioral data from the IMAGEN project, which integrates brain imaging and genetics to study biological, psychological, and environmental influences on mental health. Based on prior studies (Jia et al., 2020), we selected 19 behavioral measures, including ADHD, depression, alcohol consumption, and smoking, to compute their BAVs and DmAVs (Figure1).
Results:
For different sparse networks, we need to introduce a scaling factor for calibration before calculating DmAV, achieving highly consistent DmAV estimates. The most significant DmAV values were selected as overall heritability estimates based on the DmDI sparse pattern. Observed BAVs and DmAVs closely matched expected values across all five simulation settings (Figure 2).
We calculated BAV and DmAV for behaviors in the IMAGEN cohort (age 14, n=1979; age 19, n=1132). The highest BAVs were associated with intelligence scores, while internalizing behaviors, such as depressive symptoms, had significant BAVs with gray matter volume (BAV = 0.040, pFDR < 0.001). The highest DmAVs were associated with drug intake (DmAV = 0.350, pFDR <0.001), smoking (DmAV = 0.290, pFDR < 0.001), and drinking (DmAV = 0.158, pFDR = 0.012). Some internalizing behaviors also had significant DmAVs.
We calculated BAV and DmAV for behaviors in the IMAGEN cohort (age 14, n=1979; age 19, n=1132). The highest BAVs were associated with intelligence scores, while internalizing behaviors, such as depressive symptoms, had significant BAVs with gray matter volume (BAV = 0.040, pFDR < 0.001). The highest DmAVs were associated with drug intake (DmAV = 0.350, pFDR <0.001), smoking (DmAV = 0.290, pFDR < 0.001), and drinking (DmAV = 0.158, pFDR = 0.012). Some internalizing behaviors also had significant DmAVs.
Conclusions:
We introduced a robust VDI pattern and DmDI sparse pattern regression model to assess the overall effect of brain and DNAm on behavior. Our method can be applied to structural MRI of gray matter volume and functional MRI tasks such as MID and SST to explore brain co-activation patterns. For DNAm networks, balancing block density through sparsification is crucial for estimating variance contributions. This approach reveals shared neural and genetic bases across mental disorders and cognitive behaviors, providing insights beyond symptom-based analyses. We successfully applied our method to the IMAGEN project, offering a significant advancement in neuroimaging research.
Disorders of the Nervous System:
Neurodevelopmental/ Early Life (eg. ADHD, autism)
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 2
Emotion, Motivation and Social Neuroscience:
Emotional Perception
Genetics:
Genetic Modeling and Analysis Methods 1
Genetics Other
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural)
fMRI Connectivity and Network Modeling
Keywords:
Addictions
Affective Disorders
Anxiety
Data analysis
Development
DISORDERS
Emotions
fMRI CONTRAST MECHANISMS
MRI
Structures
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
Provide references using APA citation style.
[2] Eklund, A., Nichols, T. E., & Knutsson, H. (2016). Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates. Proceedings of the National Academy of Sciences of the United States of America, 113(28), 7900–7905. https://doi.org/10.1073/pnas.1602413113
[3] Gong, W. K., Wan, L., Lu, W. L., Ma, L., Cheng, F., Cheng, W., Grünewald, S., & Feng, J. F. (2018). Statistical testing and power analysis for brain-wide association study. Medical Image Analysis, 47, 15–30. https://doi.org/10.1016/j.media.2018.03.014
[4] Jia, T. Y., Ing, A., Quinlan, E. B., Tay, N., Luo, Q., Francesca, B., Banaschewski, T., Barker, G. J., Bokde, A. L. W., Bromberg, U., Büchel, C., Desrivières, S., Feng, J., Flor, H., Grigis, A., Garavan, H., Gowland, P., Heinz, A., Ittermann, B., … Consortium, I. (2020). Neurobehavioural characterisation and stratification of reinforcement-related behaviour. Nature Human Behaviour, 4(5), 544. https://doi.org/10.1038/s41562-020-0846-5
[5] Moore, L. D., Le, T., & Fan, G. P. (2013). DNA methylation and its basic function. Neuropsychopharmacology, 38, 23–38.
[6] Zhao, W. Q., Palmer, C. E., Thompson, W. K., Chaarani, B., Garavan, H. P., Casey, B. J., Jernigan, T. L., Dale, A. M., & Fan, C. C. (2021). Individual differences in cognitive performance are better predicted by global rather than localized BOLD activity patterns across the cortex. Cerebral Cortex, 31(3), 1478–1488. https://doi.org/10.1093/cercor/bhaa290