Data Quality
At the suggestion of the REMCs, EDACC now computes Variability QC metrics for each experiment. Our methodology involves the following steps:
- For each experiment, compute the average epigenomic signal over 10,000 bp windows tiling the entire genome
- Compute Pearson correlation with all available replicates, biological or technical
- Report the Pearson correlation with the closest replicate
- Report NA when no replicate is available
The Variability QC scores for Data Release 3 are available in this spreadsheet. For Data Release 4 and future releases, the Variability QC scores will be reported as part of the metadata submitted to GEO. The QC metrics already computed and reported are described at the Roadmap Epigenomics site and encompass:
- SPOT/PTIH
- FindPeaks reads enrichment
- Poisson enrichment
- iROC
EDACC performs the following three types of quality checks and measurements on submissions from REMCs:
- Compliance with EDACC Recommendations. This includes checks of compliance with agreed EDACC Recommendations for content, formatting, semantics, and metadata.
- Verification of Level 2 data. Quality metrics (See current EDACC Recommendations document for details) are collected for specific assays. Quality metrics are examined for outliers. The outliers are assigned status "verification issue" and are resolved in consultation between EDACC and the submitting REMC.
- Validation of Level 3 data. Validation involves comparison of assay results obtained on the same type of sample using different technologies or methods.
For more details consult the current version of the EDACC Recommendations document.