Pathogen Variant Surveillance

What it does

This skill pulls live genomic surveillance data for SARS-CoV-2, H5N1, seasonal influenza, RSV, mpox, measles, dengue, West Nile, Ebola and more from public GenSpectrum LAPIS instances. Its core rule is that a lineage name or prevalence figure is never produced from model memory — every number is a count returned by a live instance, stamped with the data version it came from.

Four Python scripts (standard library only):

• resolve_lineage.py — is this name still designated, what does the alias unpack to, was it withdrawn or redesignated?
• lineage_prevalence.py — weekly share with Wilson confidence intervals plus a descriptive log-odds growth slope
• mutation_profile.py — mutation profile, lineage-vs-lineage diff, nucleotide mode for primer/probe questions
• reporting_lag.py — completeness curve telling you how far back the data can be trusted

Field names are read from each instance's databaseConfig at run time rather than assumed, and the doc catalogues ten traps that otherwise produce silently wrong answers (bare lineage names excluding descendants, * wildcards behaving inversely on unindexed columns, and so on).

Who it's for

• Infectious-disease researchers who must cite lineage names and prevalence in manuscripts
• Diagnostics developers checking whether PCR primers/probes still match circulating sequence
• Epidemiology, public-health and bioinformatics analysts
• Anyone who wants to eliminate stale or withdrawn lineage names from AI output

Example uses

1. "What variant is dominant in the US right now?" — run lineage_prevalence.py --top 5 --where country=USA --weeks 12 to discover the top lineages from data instead of naming them up front.
2. Gate a manuscript's lineage list — resolve_lineage.py XFG.23.1.3 PQ.17 PC.2 flags withdrawn or redesignated names and exits 1 so CI can fail the check.
3. Assay integrity check — mutation_profile.py "XFJ*" --versus "XFG*" --nucleotide to see whether the target region has drifted.
4. Decide what's quotable — reporting_lag.py --where country=USA before citing any recent prevalence; on H5N1 the recent picture can be effectively blind for two months.

Install the skill I found on Claude Skill Mart.
Copy the skills/pathogen-variant-surveillance folder from the GitHub repo K-Dense-AI/scientific-agent-skills into my ~/.claude/skills/pathogen-variant-surveillance/.
When it's done, tell me in one line what this skill can do.