08 – Loading Reference Genome into Chado

Session Overview

This lab loads reference genome feature data directly into the PostgreSQL Chado schema using PostgreSQL's \copy command. Participants will connect to the training database and bulk-import Nipponbare reference genome features from a CSV file, then verify the load with row count queries.

What You Will Do

Connect to the chado_training database using psql
Load the Nipponbare reference genome features with \copy
Verify the loaded data with SQL queries

Where to Get a Quality Reference Genome

For rice, a high-quality annotated reference genome (GFF + FASTA) can be downloaded from RiceOme — a comprehensive rice omics database maintained by Huazhong Agricultural University:

🌎 riceome.hzau.edu.cn/download.html

The GFF annotation file from this source is compatible with both loadReferenceGenome.sh and loadReferenceGenomeWSeq.sh provided in the Related Files section.

Related Files

Download these before starting the lab:

📄 11-feature-nipponbare.csv 📄 loadReferenceGenome.sh 📄 loadReferenceGenomeWSeq.sh

Option A — Load a Pre-built CSV with \copy

Use this option when a ready-made feature CSV is already available (e.g., the Nipponbare training file). Make sure you are in the directory containing the CSV file before connecting.

Step 1: Connect to the Database

bash

sudo -u postgres psql -d chado_training

Step 2: Load the Reference Genome Features

psql

\copy feature FROM '11-feature-nipponbare.csv' WITH (FORMAT csv, HEADER true)

Step 3: Verify

psql

SELECT COUNT(*) FROM feature;

Option B — Load from GFF Using a Script

When you have a raw GFF annotation file instead of a pre-built CSV, use one of the two loader scripts. Both scripts automatically look up or insert the organism, CV, and cvterm entries, then generate intermediate CSVs and bulk-load them into Chado.

Which Script to Use?

Script	When to use	Chromosome seqlen source	Chromosome naming
`loadReferenceGenome.sh`	GFF has only `gene` rows (no explicit chromosome rows)	Inferred from the last gene position per chromosome	`Chr1`, `Chr2` …
`loadReferenceGenomeWSeq.sh`	GFF has explicit `chromosome`-type rows with real sequence lengths	Read directly from the `chromosome` row in the GFF	`chr01`, `chr02` … (zero-padded)

What the Scripts Do

Look up or insert the organism in the organism table
Look up or insert the sequence CV and chromosome / gene cvterms
Generate <output>_feature.csv and <output>_featureloc.csv
Bulk-load both CSVs into Chado using \copy

Parameters

Flag	Description
`-i`	Input GFF file
`-o`	Output file prefix (intermediate CSVs are named `<prefix>_feature.csv` etc.)
`-n`	Organism common name (must match what is in the database)
`-p`	PostgreSQL password (`training2026` for the training database)

Step 1: Make the Script Executable

bash

chmod +x loadReferenceGenome.sh

Step 2: Run the Script

bash

bash loadReferenceGenome.sh -i nipponbare.gff -o nipponbare -n "Japonica nipponbare" -p training2026

Or, if using the version with explicit chromosome lengths:

bash

bash loadReferenceGenomeWSeq.sh -i nipponbare.gff -o nipponbare -n "Japonica nipponbare" -p training2026

Step 3: Verify

psql

sudo -u postgres psql -d chado_training -c "SELECT COUNT(*) FROM feature;"
sudo -u postgres psql -d chado_training -c "SELECT COUNT(*) FROM featureloc;"

Expected Outcome

Chromosome and gene features are inserted into the feature table with correct organism and cvterm linkage.
Gene locations are recorded in featureloc with chromosome feature as the source feature (srcfeature_id).
Data is queryable using SQL joins across feature, featureloc, and organism.

ⓘ Once the load completes, proceed to Session 09 — Loading Dataset into Chado to load the full training dataset using uploadDataset.sh.