E.g. Aydemir 2018 and Verity 2020 we know that Verity bundles the Aydemir samples with the additional ones in Verity 2020, so these are double counted.
This is HARD problem to fix without looking back at papers to check where samples were sourced from. One approach to identify possible papers where this will be an issue would be checking for same lat long and start year across studies and identifying repeated entries for the same mutation:
E.g. for WWARN:
clean_wwarn %>% mutate(lly = paste0(lat, long, year)) %>% split(.$lly) %>% lapply(function(x){x$issues = (length(unique(x$mut)) < nrow(x)); return(x)}) %>% do.call(rbind, .) %>% filter(issues) %>% split(.$lly)
Example possible study issue here for example:
$`9.60222-13.79162014`
iso3c admin_0 admin_1 site lat long year study_start_year study_end_year x n prev gene mut gene_mut annotation
9.60222-13.79162014.190 GIN Guinea Conakry Conakry 9.60222 -13.7916 2014 2012 2016 3 10 0.3000000 crt crt_76T crt:76:T <NA>
9.60222-13.79162014.2091 GIN Guinea Conakry Conakry 9.60222 -13.7916 2014 2012 2016 2 10 0.2000000 mdr1 mdr1_86Y mdr1:86:Y <NA>
9.60222-13.79162014.2313 GIN Guinea Conakry Conakry 9.60222 -13.7916 2014 2012 2016 19 33 0.5757576 mdr1 mdr1_86Y mdr1:86:Y <NA>
database pmid url
9.60222-13.79162014.190 WWARN 29436339 https://www.ncbi.nlm.nih.gov/pubmed/29436339
9.60222-13.79162014.2091 WWARN 29436339 https://www.ncbi.nlm.nih.gov/pubmed/29436339
9.60222-13.79162014.2313 WWARN 30259821 https://www.ncbi.nlm.nih.gov/pubmed/?term=30259821
Here it seems strange that a different pmid paper had the same lat long and sample duration but was in two different studies.
E.g. Aydemir 2018 and Verity 2020 we know that Verity bundles the Aydemir samples with the additional ones in Verity 2020, so these are double counted.
This is HARD problem to fix without looking back at papers to check where samples were sourced from. One approach to identify possible papers where this will be an issue would be checking for same lat long and start year across studies and identifying repeated entries for the same mutation:
E.g. for WWARN:
clean_wwarn %>% mutate(lly = paste0(lat, long, year)) %>% split(.$lly) %>% lapply(function(x){x$issues = (length(unique(x$mut)) < nrow(x)); return(x)}) %>% do.call(rbind, .) %>% filter(issues) %>% split(.$lly)Example possible study issue here for example:
Here it seems strange that a different pmid paper had the same lat long and sample duration but was in two different studies.