Surveillance workflow"

In lineagefreq: Lineage Frequency Dynamics from Genomic Surveillance Counts

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.width = 7,
  fig.height = 4
)

Overview

This vignette demonstrates a complete end-to-end surveillance analysis: from raw count data to actionable outputs. The workflow mirrors what a public health genomics team would run weekly.

Step 1: Load and prepare data

library(lineagefreq)

data(sarscov2_us_2022)
x <- lfq_data(sarscov2_us_2022,
              lineage = variant,
              date    = date,
              count   = count,
              total   = total)

Step 2: Collapse rare lineages

Real surveillance data often contains dozens of low-frequency lineages. collapse_lineages() merges those below a threshold into an "Other" category.

x_clean <- collapse_lineages(x, min_freq = 0.02)
attr(x_clean, "lineages")

Step 3: Fit model

fit <- fit_model(x_clean, engine = "mlr")
summary(fit)

Step 4: Growth advantages

ga <- growth_advantage(fit,
                       type = "relative_Rt",
                       generation_time = 5)
ga

autoplot(fit, type = "advantage", generation_time = 5)

Step 5: Identify emerging lineages

emerging <- summarize_emerging(x_clean)
emerging[emerging$significant, ]

Step 6: Forecast

fc <- forecast(fit, horizon = 28)
autoplot(fc)

Step 7: Assess sequencing needs

How many sequences are needed to reliably detect a lineage at 2% frequency?

sequencing_power(
  target_precision = 0.05,
  current_freq = c(0.01, 0.02, 0.05, 0.10)
)

Step 8: Extract tidy results

All results are compatible with the broom ecosystem.

tidy.lfq_fit(fit)
glance.lfq_fit(fit)

Summary

A typical weekly workflow:

1. lfq_data() — ingest new counts
2. collapse_lineages() — clean up rare lineages
3. fit_model() — estimate dynamics
4. summarize_emerging() — flag growing lineages
5. forecast() — project forward 4 weeks
6. autoplot() — generate report figures

lineagefreq documentation built on April 3, 2026, 9:09 a.m.