6.4. Where Does Data Go?

This section assumes that the reader is familiar with the Chado relational database schema. It will explain where data is stored and how different data types within the same TPPS submission are associated with one another.

6.4.1. Dbxref x Project

For every submission to TPPS, a dbxref record is created to represent the submission as a whole. The dbxref record is assigned a unique accession number (TGDRxxxx) which is used to associate all other data in the submission to this reference. The dbxref record is also associated with a project record through the project_dbxref table. It’s within the project record that metadata about the study (sequencing method, study type, data file, study lat/lon) in the projectprop table is stored.

6.4.2. Stock

For each individual plant in the submission, a stock record is created with a uniquename with the format <Project Accession>-<Plant ID>. The metadata for the plant (location) is then represented with associated stockprop records. The stock is associated with the project through the project_stock table.

6.4.3. Phenotype

For each phenotype measurement, a phenotype record is created with a uniquename with the format <Project Accession>-<Plant ID>-<Phenotype Name>. The value is the measured value of the phenotype (or null if value is not applicable). The metadata for the phenotype (unit, description, min/max values, etc) is then represented with associated phenotypeprop records. The phenotype is associated with the plant through the stock_id of the base phenotype record, as well as through the stock_phenotype table.

Here is a representation of the relationships between some of the phenotype tables for reference:

6.4.4. Genotype

How genotype data gets submitted in TPPS varies slightly based on the type of genotype data and the format it is provided in, but the locations in Chado and the associations between the data stay almost entirely the same. The function that records all of this data can be found here.

For each genotype marker, two separate feature records are created, called the marker and variant records. In addition to the feature records, a genotype record is created. Each genotype call for each individual is represented in the genotype_call table, which associates the marker, variant, genotype, stock, and project together and also holds any additional metadata about the call. Each stock is also associated with each genotype in the stock_genotype table.

If association data is available for a genotype marker, then an additional feature record is created with a feature_cvterm for the cvterm of the associated trait and a featureprop for the type of association statistics being used (P-value, Bonferroni, etc). A feature_relationship record will then be created between the association feature and the variant feature, and the value of this record will be the calculated association confidence value.

Here is a representation of the relationships between some of the genotype tables for reference: