Characterize influenza sequences from Madagascar and Africa available in GISAID
# Madagascar Influenza
*Last updated: 2026-04-23*
**Report:**
## Contents
- Goal
- Data
- Alignment pipeline
- Phylogenetic tree pipeline
- Reproducing the analysis
- Progress
## Goal
Do Malagasy influenza viruses form distinct local lineages (monophyletic clades), or are they sporadic
importations from the broader African gene pool?
This determines whether sustained in-country surveillance in Madagascar is justified from a genomic-epidemiology perspective.
## Data
### GISAID (downloaded 16 March 2026)
**30,065 isolates** (all Africa, all hosts, after deduplication) — **1,849 from Madagascar (6.1%)**.
Raw GISAID files are not tracked by git; derived outputs (`combined_metadata.tsv`) are.
| Tag | Description | Isolates | Sequences |
|-----|-------------|----------|-----------|
| `a_human_africa` | Flu A human Africa | 19,902 | 110,901 |
| `b_human_africa` | Flu B human Africa | 5,081 | 27,900 |
| `animal_environmental_africa` | All animal/env Africa | 5,160 | 22,314 |
| **Total** | after deduplication | **30,065** | **161,143** |
**782 / 1,849 Madagascar isolates (42.3%) are analysis-ready** (complete 8-segment genome + known location + known date).
### Norosoa Razanajatovo (unpublished avian, 2021–2023)
109 avian isolates, 830 segment sequences from Institut Pasteur de Madagascar.
Files in `data/norosoa_avian_*.fasta` — not tracked by git.
| Subtypes | Isolates |
|----------|----------|
| H9N2 | 55 |
| H4 (partial) | 14 |
| H6N2, H6N8 | 13 |
| H1N2 | 8 |
| other | 19 |
Notes: 3 isolates already in GISAID (partial); 1 confirmed co-infection (H1N2/H6N1); 1 confirmed import from Brazil (H4N6).
## Alignment pipeline
**358 MAFFT alignments** — one per segment × subtype × scope (Africa / Madagascar), filtered to ≥ 3 sequences.
Input: 161,143 GISAID + 830 Norosoa sequences. Outputs: `alignments/aligned/`.
```bash
bash scripts/align_segments.sh # submits SLURM jobs (resumable)
squeue -u $USER # monitor
```
| Script | Purpo …