biocViews-dumbbell.R

## package biocViews dumbbell

library(BiocPkgTools)
library(dplyr)
library(tidyr)
library(graph)

library(biocViews)
data(biocViewsVocab)

first <- biocPkgList("3.7") %>% select(Package)
second <- biocPkgList("3.9") %>% select(Package, biocViews) %>% unnest()
second <- second %>%
    mutate(New = ifelse(Package %in% first$Package, "Existing", "New"))

views <- second %>% group_by(biocViews, New) %>% count() %>%
    spread(New, n) %>% arrange(desc(New))

new_pkgs <- anti_join(second, first, by="Package")

edges <- tibble(edges = edgeNames(biocViewsVocab)) %>%
    separate(edges, c("from", "to"), sep="~")

flds <- c("ResearchField", "As.Character", "Technology", "Sequencing")

ppn <- left_join(edges, views, by = c(to="biocViews")) %>%
    filter(from %in% flds, !(is.na(Existing) & is.na(New))) %>%
    replace_na(list(Existing = 0, New = 0)) %>%
    arrange(from, desc(New)) %>%
    group_by(from) %>%
    mutate(Existing = Existing / sum(Existing), New = New / sum(New)) %>%
    ungroup()

rsrch_fld <- ppn %>% filter(from == "ResearchField") %>%
    mutate(to = factor(to, levels = to))

assay_domain <- ppn %>% filter(from == "AssayDomain") %>%
    mutate(to = factor(to, levels = to))

tech <- ppn %>% filter(from == "Technology") %>%
    mutate(to = factor(to, levels = to))

sequencing <- ppn %>% filter(from == "Sequencing") %>%
    mutate(to = factor(to, levels = to))

library(ggplot2)
library(ggalt)

dumbbell <-
    function(df, ylab, pdf = TRUE)
{
    p <- ggplot(df, aes(x = Existing, xend = New, y = to)) +
        geom_dumbbell(colour="#a3c4dc", size=4, colour_xend="#0e668b") +
        labs(title = ylab, x = "Proportion of packages", y = NULL) +
        xlim(0, 0.6)
    if (pdf) {
        pdf(
            paste0(gsub(" +", "-", tolower(ylab)), "-dumbbell.pdf"),
            height = 5
        )
        print(p)
        dev.off()
    } else p
}

dumbbell(rsrch_fld, "Research field")
dumbbell(assay_domain, "Assay domain")
dumbbell(sequencing, "Sequencing method")
dumbbell(tech, "Technology")