Brain Genomics Superstruct Project (GSP)
Large-Scale Human Neuroimaging, Behavioral, and Cognitive Dataset
The Brain Genomics Superstruct Project (GSP) is a large-scale open dataset designed to enable the study of relationships between brain structure, brain function, behavior, and genetics. It provides a carefully curated and quality-controlled dataset of over 1,500 healthy human participants, integrating multimodal MRI with extensive phenotypic data.
- Source: https://www.neuroinfo.org/gsp
- Primary reference: Holmes et al., 2015, Scientific Data :contentReference[oaicite:0]{index=0}
Dataset Overview
-
Participants: 1,570 healthy individuals
-
Age range: 18–35 years
-
Population: Non-clinical (screened for neurological and psychiatric conditions)
-
Modalities:
- Structural MRI (T1-weighted)
- Resting-state fMRI (1–2 runs per subject)
-
Additional data:
- Behavioral measures
- Cognitive assessments
- Personality traits
- Demographics and health questionnaires
- (Planned/partial) genetic data
Study Design
The GSP was designed as a high-throughput, large-scale aggregation effort, leveraging existing studies across the Boston research community:
- Standardized MRI protocol applied across multiple sites
- Short acquisition time (~15–30 minutes total)
- Integration with behavioral and cognitive testing
- Linked imaging with phenotypic and genetic data
This strategy enabled rapid accumulation of thousands of datasets within a few years :contentReference[oaicite:1]{index=1}
MRI Acquisition
Structural MRI
- Sequence: Multi-echo MPRAGE
- Resolution: ~1.2 mm isotropic
- Optimized for fast acquisition (~2 minutes)
Resting-State fMRI
- Sequence: Gradient-echo EPI (BOLD)
- TR: 3000 ms
- Volumes: 120 timepoints (after stabilization)
- Whole-brain coverage including cerebellum
Participants were instructed to:
- Keep eyes open
- Stay awake and still
Data Quality and Processing
-
Automated quality control applied to all scans
-
Exclusion criteria included:
- Motion artifacts
- Low temporal SNR
- Anatomical abnormalities
-
Provided quality metrics:
- Temporal signal-to-noise ratio (tSNR)
- Motion parameters
- Functional data quality indices
-
Data format:
- Imaging: NIfTI
- Phenotypic data: CSV
Behavioral and Cognitive Measures
A large subset of participants completed:
- Personality assessments (e.g., neuroticism, anxiety)
- Cognitive tasks (e.g., mental rotation, attention tasks)
- Self-report behavioral measures
These measures enable:
- Brain–behavior association studies
- Individual difference analyses
- Cognitive neuroscience modeling
Special Features
Large Sample Size
- Enables detection of small effect sizes in brain–behavior relationships
Test–Retest Subset
- 69 participants scanned twice
- Supports reliability and reproducibility studies
Precomputed Outputs
- Morphometric measures
- Functional connectivity matrices
- Quality metrics
Scientific Applications
The GSP dataset supports:
- Functional connectivity and network analysis
- Brain–behavior relationship studies
- Cognitive and personality neuroscience
- Method development and benchmarking
- Reliability and reproducibility research
Limitations
- Convenience sample (Boston area, highly educated population)
- Short scan duration limits some analyses
- Demographic distribution not fully representative
Usage Agreement
Data access requires agreement to the official terms:
Typical requirements include:
- Use for scientific research only
- Proper citation of the dataset
- Compliance with data use restrictions
Data Access
Available through:
- Harvard Dataverse
- LONI Image Data Archive
Users must request access and comply with data use terms.
Citation
Holmes AJ et al. (2015)
Brain Genomics Superstruct Project initial data release with structural, functional, and behavioral measures
Scientific Data 2:150031