Additional inference howto posts

[adamkucharski]

• Please check if the PR fulfills these requirements

• I have read the CONTRIBUTING guidelines
• A new item has been added to NEWS.md – not available in repo?
• Tests for the changes have been added (for bug fixes / features) – NA as howto script
• Docs have been added / updated (for bug fixes / features) – NA as howto script
• Checks have been run locally and pass– NA as howto script

• What kind of change does this PR introduce? (Bug fix, feature, docs update, ...)
  This is a PR with 3 new howto guides:

• MCMC inference of R and k from cluster sizes for MERS using epichains and MCMCpack: reproduction_number_cluster_size.qmd
• Reconstruction of infection curves for COVID and Ebola using EpiNow2: estimate_infections.qmd (can also create PR to EpiNow2 to add this as a case study link on their page once merged here)
• Inference of R for H1N1 pandemic influenza from serological data using finalsize and socialmixr: reproduction_number_serological_data.qmd

[avallecam]

Thanks for the contribution! I'll rebase this PR and force-push some commits to see how the checks react

[avallecam]

Thanks for the contribution!

The reproduction_number_cluster_size.qmd and reproduction_number_serological_data.qmd work fine! However, the estimate_infections.qmd is not running, which limited further review. This seems to be due to reading the URL for the UK COVID dashboard API, which may be blocking the building of the whole website.

Could we verify if this is still working for you? If this needs preliminary configurations to run, should we try an alternative dataset or reading approach?

[avallecam]

I currently get this output from uk_data object. Could you verify if you still get a data frame output?

> uk_data
  [1] X..DOCTYPE.html..html.lang.en.class.govuk.template.__..(...)   
  [2] queries...state..data..success.true    
  [3] data...status.Green    
  [4] geography_name.North.East    
  [5] geography_code.E12000001    
  [6] refresh_date.null.   
  [7] X.status.Green    
  [8] geography_name.North.West    
  [9] geography_code.E12000002    
 [10] refresh_date.null..1    
 [11] X.status.Green.1    
 [12] geography_name.Yorkshire.and.The.Humber    
 [13] geography_code.E12000003    
 [14] refresh_date.null..2    
 [15] X.status.Green.2    
 [16] geography_name.East.Midlands    
 [17] geography_code.E12000004    
 [18] refresh_date.null..3    
 [19] X.status.Green.3    
 [20] geography_name.West.Midlands    
 [21] geography_code.E12000005    
 [22] refresh_date.null..4    
 [23] X.status.Green.4    
 [24] geography_name.East.of.England    
 [25] geography_code.E12000006    
 [26] refresh_date.null..5    
 [27] X.status.Green.5    
 [28] geography_name.London    
 [29] geography_code.E12000007    
 [30] refresh_date.null..6    
 [31] X.status.Green.6    
 [32] geography_name.South.East    
 [33] geography_code.E12000008    
 [34] refresh_date.null..7    
 [35] X.status.Green.7    
 [36] geography_name.South.West    
 [37] geography_code.E12000009    
 [38] refresh_date.null... 
 [39] dataUpdateCount.1    
 [40] dataUpdatedAt.1717579083479    
 [41] error.null    
 [42] errorUpdateCount.0    
 [43] errorUpdatedAt.0    
 [44] fetchFailureCount.0    
 [45] fetchFailureReason.null    
 [46] fetchMeta.null    
 [47] isInvalidated.false    
 [48] status.success    
 [49] fetchStatus.idle.   
 [50] queryKey..weather.alert.list    
 [51] heat.   
 [52] queryHash...weather.alert.list...heat..

[avallecam]

If we want to prioritize reusing existing data, similar data may be available in {incidence2} and {cfr}

incidence2::covidregionaldataUK contains one-year data for the UK, and cfr::covid_data contains three-year data for 19 countries, including the UK.

library(tidyverse)

incidence2::covidregionaldataUK %>% 
  # preprocess missing values
  tidyr::replace_na(list(deaths_new = 0)) %>%
  # compute the daily incidence
  incidence2::incidence(
    date_index = "date",
    counts = "deaths_new",
    count_values_to = "confirm",
    date_names_to = "date"
  ) %>%
  # complete range of dates
  incidence2::complete_dates()
#> # incidence:  490 x 3
#> # count vars: deaths_new
#>    date       count_variable confirm
#>    <date>     <fct>            <dbl>
#>  1 2020-01-30 deaths_new           0
#>  2 2020-01-31 deaths_new           0
#>  3 2020-02-01 deaths_new           0
#>  4 2020-02-02 deaths_new           0
#>  5 2020-02-03 deaths_new           0
#>  6 2020-02-04 deaths_new           0
#>  7 2020-02-05 deaths_new           0
#>  8 2020-02-06 deaths_new           0
#>  9 2020-02-07 deaths_new           0
#> 10 2020-02-08 deaths_new           0
#> # ℹ 480 more rows

cfr::covid_data %>% 
  # compute the daily incidence
  incidence2::incidence(
    date_index = "date",
    counts = "deaths",
    groups = "country"
  ) %>% 
  # complete range of dates
  incidence2::complete_dates()
#> # incidence:  20,786 x 4
#> # count vars: deaths
#> # groups:     country
#>    date_index country   count_variable count
#>    <date>     <chr>     <fct>          <dbl>
#>  1 2020-01-03 Argentina deaths             0
#>  2 2020-01-03 Brazil    deaths             0
#>  3 2020-01-03 Colombia  deaths             0
#>  4 2020-01-03 France    deaths             0
#>  5 2020-01-03 Germany   deaths             0
#>  6 2020-01-03 India     deaths             0
#>  7 2020-01-03 Indonesia deaths             0
#>  8 2020-01-03 Iran      deaths             0
#>  9 2020-01-03 Italy     deaths             0
#> 10 2020-01-03 Mexico    deaths             0
#> # ℹ 20,776 more rows

^{Created on 2024-06-06 with reprex v2.1.0}

[TimTaylor]

@avallecam - Cool. FWIW - In version 2.3.0 I've also added an argument to complete_dates within the function call itself. If useful you can also use named vectors of the form c(new_name = "old_name") to rename on the fly. E.g. for the first example you could do

incidence2::covidregionaldataUK %>% 
  # preprocess missing values
  tidyr::replace_na(list(deaths_new = 0)) %>%
  # compute the daily incidence renaming deaths_new and completing dates
  incidence2::incidence(
    date_index = "date",
    counts = c(deaths = "deaths_new"),
    count_values_to = "confirm",
    date_names_to = "date",
    complete_dates = TRUE
  )

[avallecam]

elegant! thanks for sharing the new features

[avallecam]

When I try to access this links, I get immediately redirected to https://ukhsa-dashboard.data.gov.uk/

Should we try an alternative way? or provide more details on how to access it using the API?

[avallecam]

In this first review, I propose edits to estimate_infections.qmd to make this able to run. Some additional edits not added in this review are:

• Edit also the index.qmd to make this one listed on the main page.
• Use {lintr} to adhere to the tidyverse style guides, even though we do not have the workflow in this repo. This will add to the content readability.
• Add the chunk corresponding to get one copy and paste chunk in each entry #51 for single chunks ready to copy and paste by users

If you prefer, for best use of our time, I can take charge of this PR, make the edits, and push them directly as commits. (I needed to make them locally to proof that they work)

One suggestion change I want to emphasize is the plotting. One alternative is adding layers from the EpiNow2 outputs. If not, we can use ggplot2 for a custom plot from the output. Any strong opinions about this?

[adamkucharski]

Thanks for those comments! I've committed the edits. I haven't yet added the following, as wasn't sure of most efficient way:

• lintr checks
• code to copy chunks

Feel free to add commits on these, then can merge!

[avallecam]

Nice! about your questions:

• lintr checks

We can do this by installing the {lintr} package, and if using Rstudio, we can run an Rstudio addin called "Lint current file" displaying the Rstudio Markers pane with edit suggestions.

• code to copy chunks

Yes, sorry the link was not specific. I meant this 4-lines chunk after the yaml

Feel free to add commits on these, then can merge!

Noted. We'll do this after next week

[avallecam]

this is ready to merge. pending change is to update renv.lock to latest {epiparameter} version. but keeping this to do after merging this PR. now in issue #75

[avallecam]

requested changes added. format now match required format output. The entry reproduction_number_serological_data.qmd may require some object renaming in data_summary and summary_data to clearly differentiate between the observed and estimated values. But we can do this in a separate PR too.