marianschmidt / msspchelpr

R helper functions for Second Primary Cancer analyses

msSPChelpR

What is msSPChelpR?

This package is a collection of helper functions for analyzing Second Primary Cancer (SPC) data, including functions to reshape data, to calculate patient states and analyze cancer incidence.

Installation

You can install the released version of msSPChelpR from CRAN with:

install.packages("msSPChelpR")

And the development version from GitHub with:

# install.packages("remotes")
remotes::install_github("marianschmidt/msSPChelpR")

How to use the package

An introductory vignette on the workflow of the package can be found here. See vignette("introduction").

Citation

In case you want to cite this package, please use citation('msSPChelpR') for reference information.

msspchelpr's Issues

add function recalculate_futime

[sir_byfutime] Check for t_icdcat = NA and handle missing Location of SPC correctly

NAs should appear in final results table. Expected would be 0.

add variable sorting by id to reshape_wide_dt function

Replace gender by sex where biological sex is used

using `reshape_wide_dt()` after `renumber_time_id_dt()` results in error

.shallow(x, cols = cols, retain.key = TRUE) :
can't set ALTREP truelength

Check if some unclean version of data is returned by _dt() functions or whether key needs to be reset at beginning.

add one line of total at top of results table for ir_crosstab

Check functions *_byfutime for bug where pyar can temporarily be negative and thus sum(i_pyar) is wrongly calculated

Check functions *_byfutime for bug where pyar can temporarily be negative and thus sum(i_pyar) is wrongly calculated

--> may apply to crosstab_byfutime

(function sir_byfutime was developed after bug was identified)

add progress bar to long running functions (e.g. _byfutime)

[summarize_sir_results] activate additional options / check for correct functioning

• add_total_fu

• add_total_row

• output_information

• output = "nested"

in reshape_wide after rbind_rows check for duplicate PSEUDOPATIDs and merge rows

Include data checks with assertr package

https://docs.ropensci.org/assertr/

in pat_status implement check, add FU date to labelling and reset status depending on DDIMP and date of birth

new tidyr release - make all functions compatible

tidyr release: check all functions whether they use nest() or unnest(); check whether existing functions can profit from new vignette("rectangle")

add option ref_population = TRUE to ir_crosstab

implement .data$ pronoun and make sure there is no unmet dependencies

implement option 'inclusion_restrictions' to asir function

Currently, the asir function assumes that for all regions the same period of data registration was applied. However, in case you included regional registries that have collected data for different time periods, this should be defined in the function call and the calculations should take this into account.

programm shiny dashboard "case lookup" to show all incident cancers per ID (multiple IDs should be possible)

add function reset_spc_variables

new function: create_refrates

[summarize_sir_results] reshaping into wide format does not fully work when using output = wide and both total row and total fu

ztest_dev %>% 
summarize_sir_results_dev(.,
                          summarize_groups = c("age", "region", "sex", "year"),
                          summarize_site = FALSE,
                          output = "wide",
                          output_information = "minimal",
                          add_total_row = "start",
                          add_total_fu = "start",
                          collapse_ci = FALSE,
                          shorten_total_cols = FALSE,
                          fubreak_var_name = "fu_time",
                          ybreak_var_name = "yvar_name",
                          xbreak_var_name = "none",
                          alpha = 0.05)

add option to remove empty columns in results_df for ir_crosstab

Add labelling to result dataframe returned by asir function

[calc_futime] finalize checks of options and expected errors in function implementation

finalize checks of options and expected errors in function implementation

[renumber_time_id] add check for chronology in time_id_var?

add function to calculate absolute incidence rates for summarized groups ir_abs

remove "cohort" option from sir_byfutime and create a new function to calculate refrates_df

• remove "cohort" option from sir_byfutime
• implement new function create_refrates_df

check that all changes in calculations for asir_results with futime_src="refpop" are also reflected in "cohort" calculations

check that all changes in calculations for asir_results with futime_src="refpop" are also reflected in "cohort" calculations (first half of function)

add option for percentage column after n_base for ir_crosstab

[sir function] replace t_icdcat with icdcat only (therefore rates dataset needs to be changed and all dependencies)

replace icdcat_var with site_var in all functions

[sir functions] add option to match reference rates with category "Total" for aggregated SIR over all icdcats

[summarize_sir_results] test option "xbreak_var_name"

add function ir_crosstab_futime as special version that replicates Barclay 2019 Table 1

make a new function ir_crosstab_futime and include this new function to be optionally called from ir_crosstab

Get Zenodo DOI

https://guides.github.com/activities/citable-code/

Does it make sense that summarize_group = c("icdcat") in asir function?

so far summarize_group = c("region", "year", "sex") have been implemented as they are distinct and we can simply add up observed counts and PYARs.
Does it make sense to summarize asir over multiple icdcat and how could that be implemented into asir function?

create new filter_long function

Function should be constructed as

filter_long(long_df, keep_cond="", reverse = FALSE){
1 create new dummy variable keep that is 1 if keep_cond == TRUE
2 filter based on keep == 1 or on keep= 0, if reverse ==TRUE
}

keep_condition could be tidyverse filter expression (capture 3 dots?)

Remove duplicate _tt and _dt versions where speed advantage of a single version is clear

Create sample dataset

[reshape_wide_tt] function fails, if data.table is used instead of data.frame

[sir function] check if stratification by age is correctly handled when summarizing rates and compare to melanoma example

solve conflict with caused by := between tidyverse and data.table

[pat_status] old tidyverse function produces bug on zfkd lung cancer data

• check if it still occurs with as_tibble()
• check what changes in pat_status_tt() have not been adapted in tidyverse variant [if changes are too severe, consider replacing with tidytable function

Error code

kein nicht-fehlendes Argument f�r min; gebe Inf zur�ckkein nicht-fehlendes Argument f�r max; gebe -Inf zur�ckCan't set value labels for "p_status". Infinite value range.

[vignette] remove library(tidytable) from vignette after tt 0.5.6 release

Prepare for dplyr 1.0.0

group_by --> change add to .add

mutate_at/_if/_all --> use new dplyr::across function

see https://www.tidyverse.org/blog/2020/03/dplyr-1-0-0-is-coming-soon/

Checklist

• bind_cols() should not be expected to convert to tibble
• n() and row_number() need dplyr:: prefix
• all.equal.tbl_df() should not be used
• funs() should be replaced by list()
• group_by(add = ) needs to be replaced by .add
• sample_n() and sample_frac() need to be replaced by slice_sample(n = ) and slice_sample(prop = )
• mutate_*, group_by_*, select_* and rename_* sections need to bee rewritten using across()

crosstab_byfutime: Change futime_breaks to use year breaks instead of months

check if year futime_breaks are correctly handled in all functions

[summarize_sir_results] check for correct summary if ybreak_var_name = "none"

pat_status_dt does not restore values of lifevar_dat when use_lifedatmin is enabled

new function to recode tumor id

aim give a new sequence in numbering tumor ids per patient in long dataset version

• option 1a: consecutive numbering of all primary tumors (SEER-definition) in long dataset irrespective of predating malignancies not covered by the registry
• option 1b: consecutive numbering of all primary tumors (IARC-definition) in long dataset irrespective of predating malignancies not covered by the registry

[summarize_sir_results] fix latest errors

• output = "wide" does not work with add_total_row (fu_time_sort needs to be filled before reshaping)

• check that add_total_row works with add_total_futime

[pat_status] give information for how many cases lifedatmin option was applied