Tutorial 3 Updating Vaccine Strains

Tutorial 3: Updating Vaccine Strains from GISAID

Follow along! You'll download the current seasonal vaccine strains from GISAID, reformat their headers, and save them as source/vaccine.fasta, so vaccine strains appear as reference points in every tree. About 30 minutes the first time; usually done by the maintainer when WHO announces new vaccine strains.

You need your own GISAID account (credentials are individual and never shared). GISAID data may never be committed, shared publicly or given to anyone without their own GISAID access (Data Protection).

flowchart LR
    A["Log in to GISAID"] --> B["Vaccine Reference<br/>Sequences"]
    B --> C["Copy all segments<br/>to a workset"]
    C --> D["Download FASTA"]
    D --> E["Reformat every<br/>header (7 fields)"]
    E --> F["Check, save as<br/>source/vaccine.fasta"]

Loading

The header reformatting in the fifth box is the step that matters most.

Step 1: Log in to GISAID

Go to the GISAID website and log in with your individual credentials.

Step 2: Create a workset (optional, recommended)

Create a GISAID workset named:

seasonal influenza vaccine strains

A dedicated workset makes it easier to track which vaccine strains are selected, download all their segments together, update vaccine.fasta in future cycles, and avoid mixing vaccine references with clinical sequences.

Step 3: Open the vaccine reference sequences

Within GISAID, navigate to Vaccine Reference Sequences (seasonal influenza vaccine reference strains).

Step 4: Add each strain's segments to the workset

For each vaccine strain to include:

1. Click the strain name or accession number.
2. In the strain menu, find the Sequence section.
3. Select all available segments.
4. Click Copy to and add them to seasonal influenza vaccine strains.

(Downloading segments one at a time also works, but the workset reduces the chance of missing one.) Influenza has 8 segments: select all available ones so the strain appears in every segment build.

Step 5: Download the FASTA

Open the workset, confirm the expected strains and segments are there, and download the sequences in FASTA format.

Step 6: Reformat every header ⚠️

Caution

This is the most important step. Every header must be changed to this exact 7-field format before the pipeline runs. Otherwise strains fail to load into fludb, get wrong metadata, or silently drop out of the builds.

Make each header look like this (spaces around | are fine):

>A/Wisconsin/67/2022-E2/E3 | EPI_ISL_XXXXXXXX | 2022-03-15 | E2/E3 | 4 | A | H1N1
ATG...TAA

That is: isolate name with passage history appended, isolate ID, collection date, passage history again, segment number, type, lineage. Use the real values from GISAID for each sequence (the example is illustrative). Field-by-field rules and why passage appears twice: Step 1: vaccine.fasta headers.

Important

Do not remove passage history from the strain name: field 1 must be Isolate name-Passage details/history.

Step 7: Check and save

Before saving, check that:

• Every record has a header starting with >.
• Every header has exactly 7 fields separated by |, in the required order.
• Field 1 has isolate name + passage; field 4 has passage.
• Segment numbers, type (A or B) and lineage (H1N1, H3N2, Victoria) are present.

A quick automatic check (prints nothing when all headers have 7 fields):

grep '^>' vaccine.fasta | awk -F'|' 'NF != 7'

Save the file as source/vaccine.fasta. Replace an existing one only after confirming the new file contains every vaccine strain and segment needed for the current build.

Step 8: Confirm the pipeline can see it

ls source/vaccine.fasta
grep "^>" source/vaccine.fasta

Record the download date, workset name, the EPI_SET ID GISAID gives for the selection, and your name in source/README.md (kept with the data, never committed; provenance). Then run the pipeline (Step 2). Vaccine records are loaded by fludb/scripts/upload_vaccine.py (rule upload_vaccines in ingest.smk).

You're done ✅

Vaccine strains now appear in your trees (look for the vaccine markers in Auspice).

Troubleshooting / FAQ

• A strain is missing from the trees. Check .run/logs/upload_vaccines/ for "Skipped record" lines: usually a header with too few fields, a bad segment number, or an unrecognised type/lineage.
• Yamagata strains: they load into fludb but there is no Yamagata build.
• Can I send my vaccine.fasta to a labmate? Only if they have their own GISAID account; better, they download their own.
• Which strains are current? The WHO recommendation for the season; the selection dates used in trees are in config/{subtype}/vaccine.json.

Next step → Step 2: Running the Pipeline