BGPT: Paper Review: Mouse Phenome Database

Explore by Goal

Quick Explanation Copied

Mouse Phenome Database (MPD) — what this update gives you

MPD (phenome.jax.org) is a curated repository integrating quantitative mouse phenotypes with gene expression and genotype, plus a web interface for standardized querying and visualization—here emphasizing new ontology-driven navigation and new tools for Collaborative Cross (CC) and Diversity Outbred (DO) data, alongside a refreshed SNP/genotype variation query workflow.

Long Explanation

Paper Review (Visual): Mouse Phenome Database (MPD)

Publication: 15 Nov 2013 • Journal: Nucleic Acids Research • DOI: 10.1093/nar/gkt1159

1) What the paper claims (structured)

MPD integrates quantitative phenotype data, gene expression data, and genotype data into a common annotated framework, with web-based analysis/visualization.
Scale in this snapshot: >1330 strains and >3500 phenotype measurements, plus large genotype variant coverage (SNPs/indels/SVs) and genotype-expression data summaries.
Update focus: homepage/search/navigation revamp, improved search & visualization tools, phenotype annotation to public ontologies, ontology browser, and a new SNP/genotype variation query with higher density coverage; plus new CC/DO visualization tooling and downloadable data tables.
Data acquisition bias noted: original emphasis on inbred baseline phenotypes; at the time of the report, mutant/manipulated/outbred quantitative data were less represented, with a shift toward CC/DO panels.

2) Figures from the paper’s reported summary tables

Figure A — MPD content categories (Table 1)

The CC/RI/inbred/F1/chromosome-substitution/other percentages are reported in Table 1; DO is stated separately and is excluded from the percent breakdown.

Figure B — Baseline vs treatment/intervention measurements (Table 1)

Reported as 68% baseline vs 32% treatment/intervention measurements in Table 1.

Figure C — Genotype representation scale (Table 1)

All quantities plotted are taken directly from Table 1 (numeric parts correspond to the units stated in the table).

3) Skeptical critique (what’s strong vs what’s still unclear)

Known strengths (high confidence, directly stated)

Ontology annotation + ontology browser: MPD applies value-added annotations and uses VT/MP/MA ontologies to standardize trait descriptions, with an ontology navigator for locating measurements and connecting to external resources.
CC/DO visualization tooling: MPD introduces CC visualization that renders mean/SEM across strains with founder highlighting and DO visualization that shows distribution histograms with founder means/SEM and interactive tables.
Revamped SNP/genotype variation workflow: The paper describes replacing the prior SNP functionality with a new genotype variation resource and enhancing the SNP/genotype query process with region selection, strain filtering, and links to raw data and dbSNP via rs numbers.

Uncertainties / potential blind spots (skeptical)

Representational bias: The authors explicitly acknowledge the historical emphasis on inbred baseline phenotyping and that mutant/manipulated/outbred quantitative data were not yet abundant at the time, with a shift toward CC/DO panels.
Heterogeneity of contributing studies: MPD collects and integrates strain survey data from public databases and community contributions; such sources can vary in protocols, platforms, and annotation conventions, and the paper indicates these challenges particularly for gene expression (platform mixture, low sample sizes, shifting annotation landscape).
Ontology mapping risk: Ontologies help link phenotypes across studies, but mapping phenotype measurements into controlled vocabularies can introduce systematic effects (e.g., granularity differences; ambiguous phenotype definitions). The paper emphasizes the ontology approach, but does not quantify ontology-mapping error rates in the provided text.
Validation gaps for derived analyses: The paper describes new visualization/statistical summaries (e.g., variability metrics, omit probesets with excessive variance; CC/DO tables with SD/SEM/CV/Z-score). However, from the provided content alone, we do not see rigorous external benchmarking of these computed summaries against independent ground truth beyond example figures.

4) How you can use MPD most safely (epistemic humility)

Treat MPD as curated measurement data, not a universal causal truth source. The paper positions MPD as a repository enabling integrative query/analysis, so conclusions should follow the specific dataset’s protocol and design.
Check which strain panel and phenotype category you’re querying. The paper’s own summary shows baseline vs intervention splits and different strain types; conclusions can change if you mix panels.
For gene expression, consider variability filtering decisions. MPD derives variability metrics and allows omitting probesets with too much variance at user discretion; that choice affects downstream inferences.

5) Paper-internal “map” (knowledge graph, text-based)

This graph summarizes the relationships explicitly described in the paper: MPD integrating phenotype/expression/genotype and implementing ontology annotation, CC/DO visualization, and a genotype query workflow through a web interface.

6) What would most disprove/reshape this MPD-centered perspective?

If ontology-based standardization were shown to systematically mis-map phenotype measurements (e.g., high false associations across terms), downstream cross-study aggregation would degrade. The provided text does not report mapping error rates, so this remains an empirical unknown.
If genotype annotation layers (dbSNP builds, reference genome assemblies) were not kept synchronized, SNP/genotype query results could become inconsistent across time; the paper states MPD is up-to-date for all builds mentioned, but independent audit would be needed.
If CC/DO visualization summary statistics were biased by sample size/heterogeneous study designs, the apparent strain-panel differences could be partly an artifact; the paper emphasizes controlled protocols within projects, but cross-project differences remain plausible.

Author-specific BGPT deep dives

Feedback:

Updated: April 16, 2026