index.qmd

---
title: "workshop vignette"
output: 
  pdf_document:
    latex_engine: xelatex
date: "2023-01-12"
self_contained: TRUE
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = F)
```
<br>

This vignette summarises the findings from the _100 days and 100 lines of code_ workshop, hosted in December 2022 by [Epiverse-TRACE](https://data.org/news/epiverse-trace-a-values-based-approach-to-open-source-ecosystems/).


## What should the first 100 lines of code written during an epidemic look like?

To answer this question, we invited 40 experts, including academics, field epidemiologists, and software engineers, to take part in a 3-day workshop, where they discussed the current challenges, and potential solutions, in data analytic pipelines used to analyse epidemic data.

### What R packages and tools are available to use during an epidemic?  
To investigate this in a similar setting to what an outbreak response team would experience, workshop participants were divided into groups, and asked to develop a plausible epidemic scenario, that included:

* A situation report, describing the characteristics of the epidemic

* A linelist of cases and contact tracing data, by modifying provided datasets containing simulated data

* A set of questions to address during the analytic process

Groups then exchanged epidemic scenarios and analysed the provided data to answer the questions indicated the previous group, as if they were a response team working to solve an outbreak.
Details about each of these outbreak scenarios and the analytic pipelines developed by the groups are summarised in this vignette.

### Simulating epidemic data

Before the workshop, a fictitious dataset was created, which consisted of a linelist and contact tracing information.

To generate linelist data, the package [`bpmodels`](https://github.com/epiverse-trace/bpmodels) was used to generate a branching process network. Cases were then transformed from the model output to a linelist format. To add plausible hospitalisations and deaths, delay distributions for SARS-CoV were extracted from [`epiparameter`](https://github.com/epiverse-trace/epiparameter).

To create the contact tracing database, a random number of contacts was generated for each of the cases included in the linelist. These contacts were then assigned a category of _became case_, _under follow up_ or _lost to follow up_, at random.

* Through this workshop, we identified the need for a tool to simulate outbreak data in a linelist format, to test analysis methods and other packages while having control over the characteristics of the test data. For this purpose, an R package is currently in progress, see [simulist](https://github.com/epiverse-trace/pitcher/issues/7).

## Scenario 1: Novel respiratory disease in The Gambia

<br>
```{r, echo=FALSE, out.width="100%"}
knitr::include_graphics('images/Scenario1smq.png')
```
<br>

### Analytic pipeline for scenario 1 (analysed by group 2)

* Data cleaning

  + [`linelist`](https://cran.r-project.org/web/packages/linelist/index.html) to standardise date format
  + [`cleanr`](https://github.com/Hackout3/cleanr) from previous Hackathon

* Delay distributions

  + [`fitdisrplus`](https://cran.r-project.org/web/packages/fitdistrplus/index.html) to fit parameteric distributions to scenario data
  + [`epiparameter`](https://github.com/epiverse-trace/epiparameter) to extract delay distributions from respiratory pathogens
  + [`EpiNow2`](https://github.com/epiforecasts/EpiNow2) to fit reporting delays
  + [`EpiEstim`](https://cran.r-project.org/web/packages/EpiEstim/index.html) / [`coarsedatatools`](https://cran.r-project.org/web/packages/coarseDataTools/index.html) to estimate incubation period of disease
  + [`epicontacts`](https://cran.r-project.org/web/packages/epicontacts/index.html)
  + [`mixdiff`](https://rdrr.io/cran/RBesT/man/mixdiff.html) 
  
* Population demographics

  + Would like to have had access to an R package similar to  [`ColOpenData`](https://github.com/biomac-lab/TRACE_ColOpenData)
  
* Risk factors of infection

  + Used [R4epi](https://www.r4epi.com) as a guide on how to create two-way tables and perform Chi-squared tests 
  
* Severity of disease

  + [`datadelay`](https://github.com/epiverse-trace/datadelay/) for CFR calculation
  + Implementation of method developed by [AC Ghani, 2005](https://pubmed.ncbi.nlm.nih.gov/16076827/) to estimate CFR
  
* Contact matching

  + [`diyar`](https://cran.r-project.org/web/packages/diyar/index.html) to match and link records
  + [`fuzzyjoin`](https://cran.r-project.org/web/packages/fuzzyjoin/index.html) to join contact and case data despite misspellings or missing cell contents
  
* Epi curve and maps

  + Used [`incidence`](https://cran.r-project.org/web/packages/incidence/vignettes/overview.html) and [`incidence2`](https://cran.r-project.org/web/packages/incidence2/vignettes/Introduction.html) for incidence calculation and visualisation
  + [`rasterR`](https://cran.r-project.org/web/packages/raster/index.html) to extract spatial information from library of shapefiles

* Reproduction number

  + [`APEestim`](https://github.com/kpzoo/APEestim)
  + [`bayEStim`](https://github.com/thlytras/bayEStim)
  + [`earlyR`](https://cran.r-project.org/web/packages/earlyR/index.html)
  + [`epicontacts`](https://cran.r-project.org/web/packages/epicontacts/index.html)
  + [`epidemia`](https://github.com/ImperialCollegeLondon/epidemia)
  + [`epiFilter`](https://github.com/kpzoo/EpiFilter)
  + [`EpiNow2`](https://github.com/epiforecasts/EpiNow2)
  + [`EpiEstim`](https://cran.r-project.org/web/packages/EpiEstim/index.html)
  + [`R0`](https://cran.r-project.org/web/packages/R0/index.html)
  + [`outbreaker2`](https://cran.r-project.org/web/packages/outbreaker2/index.html)
  
* Superspreading, by using [this](https://github.com/mrc-ide/EpiEstim/blob/master/vignettes/alternative_software.Rmd) resource

  + [`fitdistrplus`](https://cran.r-project.org/web/packages/fitdistrplus/index.html)
  + [`epicontacts`](https://cran.r-project.org/web/packages/epicontacts/index.html)
  
* Epidemic projections

  + [`incidence`](https://github.com/mrc-ide/EpiEstim/blob/master/vignettes/alternative_software.Rmd) R estimation using a loglinear model
  + [`projections`](https://cran.r-project.org/web/packages/projections/index.html) using Rt estimates, SI distributions and overdispersion estimates
  
* Transmission chains and strain characterisation
  + [IQtree](http://www.iqtree.org) and [nextclade](https://clades.nextstrain.org) to build a maximum likelihood tree and mannually inspect it
  + Advanced modelling through phylodynamic methods, using tools like [BEAST](https://beast.community)

<br>

Data analysis step  | Challenges 
------------- | ---------------------------------------
Data cleaning | Not knowing what packages are available for this purpose
Delay distributions | Dealing with right censoring <br> Accounting for multiple infectors
Population demographics | Lacking tools that provide information about population by age, gender, etc.
Risk factors of infection | Distinguishing between risk factors vs detecting differences in reporting frequencies among groups
Severity of disease | Knowing the prevalence of disease (denominator) <br> Right censoring <br> Varying severity of diffeent strains
Contact matching | Missing data <br> Misspellings
Epicurve and maps| NA dates entries not included <br> Reporting levels varying over time
Offspring distribution | Right censoring <br> Time varying reporting efforts <br> Assumption of a single homogeneous epidemic <br> Importation of cases
Forecasting | Underlying assumption of a given R distribution, e.g., single trend, homogeneous mixing, no saturation

<br>

## Scenario 2: Outbreak of an unidentified disease in rural Colombia

<br>

```{r, echo=FALSE, out.width="100%"}
knitr::include_graphics('images/Scenario2smq.png')
```
<br>

### Analytic pipeline for scenario 2 (analysed by group 3)

* Data cleaning: manually, using R (no packages specified), to
  + Fix data entry issues in columns _onset_date_ and _gender_
  + Check for missing data
  + Check sequence of dates: symptom onset &rarr; hospitalisation &rarr; death
  
* Data anonymisation to share with partners
  + [`fastlink`](https://cran.r-project.org/web/packages/fastLink/index.html) for probabilistic matching between cases &harr; contacts, based on names, dates, and ages
  
* Case demographics
  + [`apyramid`](https://cran.r-project.org/web/packages/apyramid/index.html) to stratify data by age, gender, and health status
  
* Reproductive number calculation, by using two approaches:
  + Manually, by calculating the number of cases generated by each source case, data management through [`dplyr`](https://dplyr.tidyverse.org) and [`data.table`](https://cran.r-project.org/web/packages/data.table/index.html)
  + Using serial interval of disease, through [`EpiEstim`](https://cran.r-project.org/web/packages/EpiEstim/index.html) or [`EpiNow2`](https://github.com/epiforecasts/EpiNow2)
  
* Severity of disease
  + Manual calculation of CFR and hospitalisation ratio
  
* Projection of hospital bed requirements
  + [`EpiNow2`](https://github.com/epiforecasts/EpiNow2) to calculate average hospitalisation duration and forecasting
  
* Zoonotic transmission of disease
  + Manual inspection of cases' occupation
  + Use of [IQtree](http://www.iqtree.org) and [`ggtree`](https://guangchuangyu.github.io/software/ggtree/) to plot phylogenetic data
  
* Superspreading
  + [`epicontacts`](https://cran.r-project.org/web/packages/epicontacts/index.html)
  
* Calculation of attack rate
  + Unable to calculate, given the lack of seroprevalence data
  
<br>

Data analysis step  | Challenges 
------------- | ---------------------------------------
Data anonymisation | Dealing with typos and missing data when generating random unique identifiers
Reproduction number | Right censoring <br> Underestimation of cases due to reporting delays
Projection of hospital bed requirements | Incomplete data (missing discharge date) <br> Undocumented functionality in R packages used
Zoonotic transmission | Poor documentation <br> Unavailability of packages in R <br> Differentiation between zoonotic transmission and risk factors- need for population data
Attack rate | Not enough information provided

## Scenario 3: Reston Ebolavirus in the Philippines

<br>

```{r, echo=FALSE, out.width="100%"}
knitr::include_graphics('images/Scenario3smq.png')
```

<br>

### Analytic pipeline for scenario 3 (analysed by group 4)

* Data cleaning
  + Importing data with [`rio`](https://cran.r-project.org/web/packages/rio/index.html), [`readxl`](https://readxl.tidyverse.org), [`readr`](https://cran.r-project.org/web/packages/readr/index.html), or [`openxlsx`](https://cran.r-project.org/web/packages/openxlsx/index.html)
  + Rename variables with [`janitor`](https://cran.r-project.org/web/packages/janitor/index.html)
  + Initial data checks with [`pointblank`](https://cran.r-project.org/web/packages/pointblank/index.html), [`assertr`](https://cran.r-project.org/web/packages/assertr/index.html), [`compareDF`](https://cran.r-project.org/web/packages/compareDF/index.html), or [`skimr`](https://cran.r-project.org/web/packages/skimr/index.html)
  + Vertical data checks with [`matchmaker`](https://cran.rstudio.com/web/packages/matchmaker/index.html), [`lubridate`](https://cran.r-project.org/web/packages/lubridate/index.html), or [`parsedate`](https://cran.r-project.org/web/packages/parsedate/index.html)
  + Horizontal data checks with [`hmatch`](https://github.com/epicentre-msf/hmatch), [`assertr`](https://cran.r-project.org/web/packages/assertr/index.html), or [`queryR`](https://github.com/epicentre-msf/queryr)
  + Detect duplicates with [`janitor`](https://cran.r-project.org/web/packages/janitor/index.html) and [`tidyverse`](https://www.tidyverse.org)
  + Checking for consistency with [`dplyr`](https://dplyr.tidyverse.org), or [`powerjoin`](https://cran.r-project.org/web/packages/powerjoin/index.html)
  + Translation with [`matchmaker`](https://cran.rstudio.com/web/packages/matchmaker/index.html)

* Delay distributions
  + [`fitdistrplus`](https://cran.r-project.org/web/packages/fitdistrplus/index.html) to fit parameteric distributions to epidemic data

* Case demographics
  + [`apyramid`](https://cran.r-project.org/web/packages/apyramid/index.html) to stratify data by age, gender, and health status
  + [`ggplot2`](https://ggplot2.tidyverse.org/reference/ggplot.html) to visualise data
  
* Outbreak description
  + [`sitrep`](https://rdrr.io/github/R4EPI/r4epi/man/sitrep-package.html) to generate reports
  
* Visualisation of geographic data
  + [`sf`](https://cran.r-project.org/web/packages/sf/index.html) for static maps
  + [`leaflet`](https://cran.r-project.org/web/packages/leaflet/index.html) for interactive maps
  
* Generation of tables
  + [`gtsummary`](https://cran.r-project.org/web/packages/gtsummary/index.html) for static tables
  + [`janitor`](https://cran.r-project.org/web/packages/janitor/index.html) for interactive tables
  
* Severity of disease
  + [`EpiNow2`](https://epiforecasts.io/EpiNow2/) and [`survival`](https://cran.r-project.org/web/packages/survival/index.html) to calculate CFR
  
* Attack rate
  + [`gadm`](https://rdrr.io/github/rspatial/geodata/man/gadm.html) to get population data
  + [`epitabulate`](https://rdrr.io/github/R4EPI/epitabulate/) to describe data
  + [`sf`](https://cran.r-project.org/web/packages/sf/index.html) and [`ggplot2`](https://ggplot2.tidyverse.org/reference/ggplot.html) to plot data
  
* Forecasting
  + [`EpiEstim`](https://cran.r-project.org/web/packages/EpiEstim/index.html)
  + [`EpiNow2`](https://epiforecasts.io/EpiNow2/)
  + [`bpmodels`](https://github.com/epiverse-trace/bpmodels)
  
* Spillover events
  + By cross-referencing contact data with occupations
  
* Effectiveness of contact tracing
  + By calculating the proportion of case follow-ups and comparing the delay of disease exposure to the follow-up delay
  
* Transmission trees
  + [`epicontacts`](https://cran.r-project.org/web/packages/epicontacts/index.html)
  + [`ggplot2`](https://ggplot2.tidyverse.org/reference/ggplot.html)
  
<br>

Data analysis step  | Challenges 
------------- | ---------------------------------------
Detection of outliers | No known tools to use
Severity of disease | Censoring
Spillover events | Missing data

<br>
 
## Scenario 4: Emerging avian influenza in Cambodia

<br>

```{r, echo=FALSE, out.width="100%"}
knitr::include_graphics('images/Scenario4smq.png')
```

<br>

### Analytic pipeline for scenario 4 (analysed by group 5)

* Data cleaning
  + [`readxl`](https://readxl.tidyverse.org) to import data
  + [`dplyr`](https://dplyr.tidyverse.org) to remove names
  + Manually scanning through excel to check for errors
* Reproduction number
  + [`EpiEstim`](https://cran.r-project.org/web/packages/EpiEstim/index.html)
* Severity of disease
  + Manually using R to detect missing cases
  + [`epiR`](https://cran.r-project.org/web/packages/epiR/index.html) to check for data censoring

<br>

Data analysis step  | Challenges 
------------- | ---------------------------------------
Data cleaning | No available R packages specific for epidemic data
Reproduction number | Difficulty finding parameter estimations in the literature
Severity | Missing cases <br> Need for an R package for systematic censoring analysis

<br>

## Scenario 5: Outbreak of respiratory disease in Canada

<br>

```{r, echo=FALSE, out.width="100%"}
knitr::include_graphics('images/Scenario5smq.png')
```

<br>

### Analytic pipeline for scenario 5 (analysed by group 1)

* Define project structure
  + Defining the script's structure with [`cookiecutter`](https://github.com/jacobcvt12/cookiecutter-R-package), [`reportfactory`](https://cran.r-project.org/web/packages/reportfactory/index.html), and [`orderly`](https://cran.rstudio.com/web/packages/orderly/index.html)
  + Ensuring reproducibility of the analysis with [iRODS](https://irods.org) and [Git](https://git-scm.com)
  + Working in a group with [GitHub](https://github.com)

* Data cleaning
  + Importing data with [`readr`](https://cran.r-project.org/web/packages/readr/index.html) or [`rio`](https://cran.r-project.org/web/packages/rio/index.html)
  + Checking for errors with [`linelist`](https://cran.r-project.org/web/packages/linelist/index.html), [`janitor`](https://cran.r-project.org/web/packages/janitor/index.html), [`parsedate`](https://cran.r-project.org/web/packages/parsedate/index.html), [`matchmaker`](https://cran.rstudio.com/web/packages/matchmaker/index.html), or [`lubridate`](https://cran.r-project.org/web/packages/lubridate/index.html)
  + [`janitor`](https://cran.r-project.org/web/packages/janitor/index.html) to eliminate duplicates
  + [`naniar`](https://cran.r-project.org/web/packages/naniar/index.html) to check for missing data
  + [`epitrix`](https://cran.r-project.org/web/packages/epitrix/index.html) to anonymise data

* Delay distributions
  + [`epitrix`](https://cran.r-project.org/web/packages/epitrix/index.html)
  + [`fitdistrplus`](https://cran.r-project.org/web/packages/fitdistrplus/index.html) to fit parameteric distributions to scenario data
  
* Case demographics
  + [`apyramid`](https://cran.r-project.org/web/packages/apyramid/index.html) to stratify data by age, gender, and health status
  
* Nowcasting
  + [`incidence2`](https://cran.r-project.org/web/packages/incidence2/vignettes/Introduction.html) to visualise incidence from linelist data
  + [`epiparameter`](https://github.com/epiverse-trace/epiparameter) to extract infectious disease parameter data
  + [`EpiEstim`](https://cran.r-project.org/web/packages/EpiEstim/index.html) or [`EpiNow2`](https://github.com/epiforecasts/EpiNow2) for Rt calculation
  
* Severity of disease
  + Calculation of hospitalisation and mortality rates- no R package specified
  
* Zoonotic transmission
  + [`forecast`](https://cran.r-project.org/web/packages/forecast/index.html)
  
* Generation of reports
  + [`incidence`](https://cran.r-project.org/web/packages/incidence/vignettes/overview.html) for static reports
  + [Quarto](https://quarto.org) and [R markdown](https://rmarkdown.rstudio.com) for dashboards

<br>

Data analysis step  | Challenges 
------------- | ---------------------------------------
Project structure | Working simultaneously on the same script and managing parallel tasks <br> Anticipating future incoming data in early pipeline design
Data cleaning | Large amount of code lines used on (reasonably) predictable cleaning (e.g. data sense checks) <br> Omitting too many data entries when simply removing _NA_ rows <br> Non standardised data formats <br> Implementing rapid quality check reports before analysis
Delay distributions | Identifying the best method to calculate, or compare functionality of tools <br> Need to fit multiple parametric distributions and return best, and store as usable objects
Severity of disease | Censoring and truncation <br> Underestimation of mild cases <br> Need database of age/gender pyramids for comparisons
Forecasts | Need option for fitting with range of plausible pathogen serial intervals and comparing results <br> Changing reporting delays over time <br> Matching inputs/outputs between packages
Zoonotic transmisison | Need for specific packages with clear documentation <br> How to compare simple trend-based forecasts