Skip to main content
Version: next

PEG Matrix Preparation

This page provides a practical guide to prepare a PEG Evidence Matrix for submission or sharing. For the exact column standards, see the PEG Evidence Matrix Standard and the column examples.

Need some background or examples? Start with the PEG Evidence Matrix Introduction and the toy example.

Before you start​

Make sure you have:

  • A single trait and a single GWAS source for the matrix → If not, consider whether your matrix should be split into multiple matrices
  • The set of variants and all candidate genes related to the variants
  • A list of evidence streams and any integration results you plan to report
  • A decision on the author conclusion integration column (if you will produce a PEG List)
  • Use a machine-readable table such as TSV
  • Do not use merged cells or styled spreadsheets
  • Keep headers stable and consistent with metadata
  • Use a consistent missing value marker (e.g., NA) if possible

Matrix structure​

A PEG matrix is a flat table where each row represents a variant and its mapped gene at a locus. This typically means:

  • The variant-gene pair is the unique identifier
  • Include all genes considered, not only top hits
  • Keep variant-centric evidence columns identical for all genes at the same variant when applicable
  • Keep gene-centric evidence columns identical for all variants mapped to the same gene when applicable

Minimum identifiers to include:

  • Primary Variant ID
  • Gene ID
  • Gene symbol

Step-by-step preparation​

1) Build the identifier columns​

  • Add the genomic identifier columns first if chromosome, position, effect allele, and other allele are represented in multiple columns
  • Use consistent gene identifiers (HGNC gene symbols, Ensembl gene IDs)
  • Confirm genome build and variant format are consistent throughout
  • Use the naming pattern Category_[stream]_[detail] where appropriate. Categories can be found in the list. When no applicable category can be used, PEGASUS suggest using Other_[CustomizedCategory]_[stream]_[detail1]_[detail2]
  • when you need multiple words in anu parts, please use CamelCase. e.g. credible set id -> CredibleSetId.
  • If multiple evidence streams are used in the same category, encode the stream in the header
  • If multiple value types are present in the same stream, distinguish them via [detail] part in the header
  • Use INT_[tag]_[detail], where [tag] is what you assigned in the metadata.
  • Ensure exactly ONE integration column is marked as the author conclusion in metadata
  • Keep integration logic transparent and documented

4) Align with metadata​

  • Every evidence and integration column must appear in PEG Metadata
  • Column headers must match exactly between matrix and metadata
  • Record sources, methods, and biosamples in metadata to make the matrix interpretable

Final checks​

  • One trait per matrix and one source GWAS per matrix
  • All genes at each locus are included
  • Column headers follow standard naming patterns
  • Missing values are consistent and clearly defined
  • Evidence and integration columns are documented in metadata

If you are unsure about any field or naming pattern, refer to the PEG Evidence Matrix Standard or contact us.