GDA Mouse

What is the Gene Expression Database (GXD) and what types of data does it contain?

The Gene Expression Database (GXD) is an extensive, well-curated community resource that collects and integrates different types of mouse developmental expression information. GXD focuses on endogenous gene expression in wild-type and mutant mice, making these data readily accessible to researchers via biologically and biomedically relevant searches. The database contains several types of expression data including:

• RNA in situ hybridization data

• Immunohistochemistry data

• RT-PCR results

• Northern blot data

• Western blot data

• RNA-Seq data

• Microarray experiment data

As of September 2020, GXD contains detailed expression data for nearly 15,000 genes, including data from numerous strains of wild-type mice and from more than 4,900 mouse mutants. The database holds over 365,000 images and more than 1.74 million expression result annotations for classical types of expression data .

How does GXD integrate with other mouse genomic resources?

GXD functions as an integral component of the larger Mouse Genome Informatics (MGI) resource, combining expression data with genetic, functional, phenotypic, and disease-oriented data. This integration enables unique and powerful search capabilities that foster insights into the molecular mechanisms of human development, differentiation, and disease.

Additionally, GXD maintains external links to expression resources from other vertebrate model organisms that are highly relevant for developmental research, including:

• Zebrafish (ZFIN)

• Xenopus (Xenbase)

• Chicken (GEISHA)

This cross-species integration allows researchers to perform comparative expression analyses across evolutionarily relevant model organisms .

How are high-throughput expression data incorporated into GXD?

GXD has expanded to include RNA-Seq and microarray data through a two-step process:

• Metadata indexing: GXD provides a searchable metadata index of mouse high-throughput expression experiments available in public repositories. The database incorporates mouse RNA-Seq and expression microarray experimental metadata from Gene Expression Omnibus (GEO) and ArrayExpress, applying GXD annotation standards to samples and attributes of experiments.

• Data integration: For RNA-Seq experiments, GXD imports uniformly processed expression data from the EBI Expression Atlas. This data undergoes quality normalization to generate transcript per million (TPM) values, which are then integrated with other expression data types in GXD .

This approach allows researchers to find comprehensive sets of high-throughput experiments using standardized search terms from controlled vocabularies and ontologies—a task that would be difficult when searching directly through repository resources due to term heterogeneity.

How can I use GXD to plan gene expression studies in mouse models of human disease?

GXD provides valuable reference data that can inform the design of gene expression studies in mouse models of human disease. When planning such studies:

• Reference baseline expression: Use GXD to establish normal expression patterns of your genes of interest across different tissues and developmental stages in wild-type mice.

• Select appropriate controls: Identify appropriate control strains by examining expression data from various mouse genetic backgrounds.

• Assess existing mutant models: Review expression data from over 4,900 mouse mutants to determine if existing models might be suitable for your research.

• Identify co-expressed genes: Use GXD's RNA-Seq data visualization tools like the Morpheus heat map to identify genes with similar expression patterns, which might function in the same pathways.

For human disease modeling specifically, GXD can help identify genes with expression patterns that correlate with disease-associated phenotypes. The database is particularly valuable given that thousands of new animal models are being created as centralized repositories for mouse models of human diseases, making GXD an essential tool for navigating this expanding landscape .

What considerations should I make when selecting RNA-Seq experiments from GXD for comparative analysis?

When selecting RNA-Seq experiments from GXD for comparative analysis, consider:

When comparing datasets, use GXD's embedded Morpheus heat map tool, which offers visualization and analysis capabilities including hierarchical clustering and nearest neighbors analysis to identify patterns across datasets.

How can I visualize and analyze RNA-Seq data within GXD?

GXD provides powerful visualization and analysis tools for RNA-Seq data through the embedded Morpheus heat map utility from the Broad Institute. This integration offers:

• Heat map visualization: Displays quantitative expression using color-coded average quantile normalized TPM values for each gene (rows) and corresponding biological replicate sample sets (columns).

• Sample annotation: Column labels in the heat map indicate anatomical structure, experiment ID, and GXD-assigned bioreplicate set ID. Stars in column labels indicate samples derived from mutant mice, distinguishing them from wild-type samples.

• Pattern recognition: Colored metadata rows below the column labels help users recognize patterns in the data, with distinct colors assigned to different metadata annotations.

• Data manipulation: Morpheus supports:

  • Primary and secondary metadata sorting

  • Additional sorting and filtering options for columns and rows

  • Hierarchical clustering

  • Nearest neighbors analysis

  • Visual enrichment tools

These capabilities allow researchers to identify expression patterns, compare experimental conditions, and generate hypotheses for further investigation.

How does GXD handle contradictory expression results between different experimental techniques?

GXD curates and presents expression data from multiple experimental techniques, allowing researchers to evaluate apparent contradictions between different methods:

• Comprehensive annotation: Each expression result is annotated with detailed experimental conditions, including:

  • Experimental technique

  • Probes/antibodies used

  • Genetic background

  • Age/developmental stage

  • Anatomical structure

• Strength indication: GXD records expression strength (present/absent or quantitative measures), allowing researchers to compare detection thresholds between techniques.

• RNA-Seq integration: The inclusion of RNA-Seq data provides a quantitative reference point, with expression values being categorized as:

  • Present: QN TPM ≥ 0.5

  • Absent: QN TPM < 0.5

This consistent approach allows for direct comparison between classical assay results and RNA-Seq data .

When encountering contradictory results, researchers should consider:

• Sensitivity differences between techniques

• Probe/antibody specificity

• Post-transcriptional regulation (explaining differences between RNA and protein detection)

• Spatial resolution limitations of different methods

What statistical approaches are recommended for differential expression analysis using GXD data?

For differential expression analysis using GXD data, researchers can employ several statistical approaches depending on the data type and research question:

For RNA-Seq data:

• Embedded Morpheus tools: GXD's integration with Morpheus provides built-in analysis capabilities including:

  • Hierarchical clustering to identify gene expression patterns

  • Nearest neighbors analysis to find related expression profiles

  • Filtering tools to focus on specific expression ranges

• External analysis: Export data for analysis in specialized software using:

  • DESeq2 for negative binomial distribution modeling

  • edgeR for empirical Bayes estimation

  • limma-voom for precision weight linear modeling

For classical expression data:

• GXD's Differential Expression Search: Use this tool to identify genes expressed in specific anatomical structures but not in others.

  • The algorithm primarily relies on positive expression results

  • Future updates will incorporate the extensive absence of expression information from RNA-Seq experiments

When combining data types, consider that:

• RNA-Seq provides more comprehensive absence of expression information

• In situ methods offer higher spatial resolution

• Different detection thresholds may require normalization steps

For optimal results, leverage GXD's consistent annotation of experimental metadata to ensure you're comparing equivalent developmental stages, genetic backgrounds, and anatomical structures.

How can I leverage GXD to identify candidate genes for disease models?

GXD offers several sophisticated approaches to identify candidate genes for disease models:

• Tissue-specific expression profiling: Use GXD's Differential Expression Search to identify genes with expression patterns restricted to disease-relevant tissues. For example, finding genes expressed in specific brain regions for neurodegenerative disease models.

• Developmental timing analysis: Examine temporal expression patterns to identify genes active during critical developmental windows related to disease onset.

• Mutant phenotype correlation: Integrate expression data with phenotypic information in the broader MGI database to identify genes whose altered expression correlates with disease-relevant phenotypes.

• Co-expression network analysis: Use Morpheus heat map clustering tools to identify co-expressed gene networks, potentially revealing functional relationships between known disease genes and novel candidates.

• Cross-species comparison: Utilize GXD's links to other model organism databases to identify evolutionarily conserved expression patterns, suggesting functional importance .

This integrated approach is particularly valuable given that the correlation between genomic responses in mouse models and human diseases can be poor. By identifying genes with expression patterns that match human disease signatures, researchers can develop more translatable mouse models for both common and chronic human diseases, including cancer, cardiovascular diseases, obesity, diabetes, and neurodegenerative conditions .

How can GXD data inform the selection and validation of mouse models for specific human diseases?

GXD data can significantly enhance the selection and validation of mouse models for human diseases through several approaches:

• Expression pattern matching: Compare expression profiles of disease-relevant genes between human patients and potential mouse models to identify models with similar molecular signatures.

• Developmental concordance: Verify that expression timing in mouse models aligns with critical developmental windows in human disease pathogenesis.

• Strain background assessment: Evaluate expression differences across mouse strains to select genetic backgrounds most appropriate for specific disease modeling.

• Validation benchmarking: Establish measurable expression-based endpoints for validating new models against established ones using standardized GXD annotation.

• Mutant characterization: Leverage GXD's extensive data from >4,900 mouse mutants to compare expression changes in your disease model against previously characterized mutants.

This approach is essential because humans differ significantly in their genetic vulnerability to common diseases, and there is often poor correlation between genomic responses in mouse models and human patients. GXD data helps address these challenges by providing a framework for selecting mouse models with expression profiles that better match human disease states .

The following table summarizes key considerations when selecting mouse models based on GXD data:

What quality control measures are applied to data in GXD?

GXD implements rigorous quality control measures throughout its data acquisition and integration processes:

• Literature curation: Systematic journal surveys identify publications with mouse developmental expression data. Each paper undergoes:

  • Initial annotation of genes, ages, and assay types

  • Detailed curation of expression patterns, experimental conditions, and genetic backgrounds

  • Use of standardized gene nomenclature and controlled vocabularies

• High-throughput data integration:

  • Experiment evaluation using supervised machine learning (linear SVM classifier) to identify in-scope experiments

  • Manual annotation of sample metadata to standardized terms

  • Consistent processing of RNA-Seq data through the EBI Expression Atlas pipeline

  • Application of uniform cutoffs (QN TPM ≥ 0.5 = Present expression; QN TPM < 0.5 = Absent expression)

• Controlled vocabularies: GXD makes extensive use of:

  • Mouse Developmental Anatomy Ontology

  • Gene Ontology

  • Mammalian Phenotype Ontology

  • Standard mouse strain and allele nomenclature

These quality control measures ensure data consistency and reliability, facilitating integration across different data types and enabling powerful cross-dataset searches that would otherwise be impossible due to term heterogeneity in primary repositories.

What are best practices for experimental design when generating data intended for submission to GXD?

When designing experiments to generate data suitable for GXD submission, researchers should follow these best practices:

• Experimental design and reporting:

  • Use standard mouse strain nomenclature (e.g., C57BL/6J, C57BL/6N) and specify substrain clearly

  • Document precise developmental stages using Theiler staging for embryos

  • Report specific anatomical structures examined using standard terminology

  • Include appropriate controls for all experimental conditions

  • Document expression in multiple tissues/structures rather than just the tissue of interest

• For in situ and immunohistochemistry studies:

  • Include whole-mount and section data when possible

  • Document probe/antibody details thoroughly

  • Image both negative and positive results with consistent magnification

  • Include counterstaining to identify anatomical structures

• For RNA-Seq studies:

  • Follow MINSEQE (Minimum Information about a high-throughput SEQuencing Experiment) guidelines

  • Include at least 3 biological replicates per condition

  • Process and store RNA samples consistently

  • Document sample metadata thoroughly, including age, sex, and precise anatomical dissection

  • Consider broader tissue sampling beyond your primary tissue of interest

• For mouse model development:

  • Document baseline expression in appropriate wild-type controls

  • Consider appropriate sample sizes based on expected effect sizes

  • For disease models, assess expression in multiple tissues to capture systemic effects

These practices ensure your data will be compatible with GXD's curation standards and maximize the value of your contribution to the research community.

How can researchers efficiently use GXD for comparative analysis of mouse and human expression data in disease modeling?

Researchers can efficiently leverage GXD for comparative analysis of mouse and human expression data through several methodological approaches:

• Ortholog mapping strategy:

  • Identify human-mouse orthologs for genes of interest

  • Compare expression patterns across species while accounting for developmental timing differences

  • Focus on conserved expression domains that suggest functional conservation

• Disease-relevant tissue prioritization:

  • Focus comparative analysis on tissues most relevant to the disease phenotype

  • Use GXD's anatomical ontology to match equivalent structures between species

  • Consider developmental trajectories when comparing embryonic tissues

• Integration with human datasets:

  • Use GXD expression data alongside human tissue atlases (GTEx, Human Protein Atlas)

  • Identify divergent expression patterns that might limit model translatability

  • Focus on conserved pathways rather than individual genes when expression patterns differ

• Validation approach:

  • Verify key expression findings in both species using comparable methods

  • Consider protein-level validation when transcript patterns differ

  • Document species-specific regulatory elements that might explain expression differences

This methodological framework is particularly valuable given that the genomic responses in mouse models often correlate poorly with human disease responses. By focusing on conserved expression patterns in disease-relevant tissues, researchers can develop more translatable mouse models for common and chronic human diseases .

How will GXD incorporate single-cell RNA-Seq data and what new research questions will this enable?

GXD is expanding to incorporate single-cell RNA-Seq (scRNA-Seq) data, which will transform research capabilities:

• Data integration plans:

  • Direct data load from GEO to supplement ArrayExpress imports

  • Complete non-redundant index of GXD-relevant RNA-Seq (bulk and single cell)

  • Proper annotation of biological source metadata for searchability

  • Integration with existing classical expression data types

• New research applications:

  • Cell-type specific expression profiling across developmental stages

  • Identification of rare cell populations relevant to disease processes

  • Developmental trajectory mapping at unprecedented resolution

  • Discovery of transient expression states missed by bulk RNA-Seq

• Methodological advancements:

  • Enhanced differential expression algorithms leveraging single-cell resolution

  • Integration of spatial information with transcriptomic data

  • Cell-type deconvolution of bulk RNA-Seq using scRNA-Seq reference data

  • Identification of cell-specific regulatory networks

These developments will enable researchers to address questions about cellular heterogeneity in development and disease, providing a more nuanced understanding of gene expression dynamics across cell types during mouse development.

How can researchers use GXD to interpret contradictory findings between different mouse models of the same disease?

When faced with contradictory findings between different mouse models of the same disease, GXD provides several analytical approaches:

• Strain background analysis:

  • Compare expression profiles across different mouse strains (e.g., C57BL/6J vs. C57BL/6N)

  • Identify strain-specific modifiers that might influence disease phenotypes

  • Use GXD's standardized strain annotations to isolate strain effects from other variables

• Methodological comparison:

  • Evaluate expression data generated by different techniques (in situ vs. RNA-Seq)

  • Consider differences in sensitivity and specificity between methods

  • Look for consistent patterns across multiple methodologies

• Developmental timing assessment:

  • Compare expression at equivalent developmental stages

  • Identify temporal differences in gene activation that might explain phenotypic variations

  • Consider age-dependent effects in postnatal disease models

• Systems biology approach:

  • Use pathway analysis to determine if different models affect the same biological processes through distinct mechanisms

  • Identify compensatory expression changes unique to specific models

  • Consider the broader expression network beyond the primary disease gene

This systematic approach can reveal why different models of the same disease produce varying results, helping researchers select the most appropriate model for their specific research questions .

By understanding the molecular basis for model differences, researchers can develop more refined hypotheses about disease mechanisms and potentially reconcile seemingly contradictory findings between models.