R: 每年发表基因简单分析

问题

研究最热门的基因是什么

分析

数据：来自@小灰侠整理，https://blog.csdn.net/coding_Joash/article/details/120579109
(白嫖使我快乐）

library(org.Hs.eg.db)

### gene id 以及对应的发表时间
gene <- read_tsv("Homo_GeneID_year.txt", col_types = cols("c", "c"))

###得到gene id对应得gene symbol
ids=toTable(org.Hs.egSYMBOL)
HS_GENE_YEAR <- inner_join(ids, gene, by=c('gene_id' = "GeneID"))

### 绘制词云图函数
### 可绘制gene symbol ，也可绘制year
plotWC <- function(keys, key_type = "year", size=0.4, shape='star'){
  if (key_type == "year"){
    gene_freq <- HS_GENE_YEAR %>%
      filter(year %in% keys) %>%
      group_by(symbol) %>%
      summarise(freq=n())
    
    wordcloud2::wordcloud2(gene_freq, size = size,shape = shape)
  }else if(key_type == "gene"){
      
      year_freq <- HS_GENE_YEAR %>%
        filter(symbol %in% keys) %>%
        group_by(year) %>%
        summarise(freq=n())
      
      wordcloud2::wordcloud2(year_freq, size = size, shape = shape)
    
  }
}

### 绘制柱状图
plot_bar <- function(genes, years=NULL){
  
  if (is.null(years)){
    years <- HS_GENE_YEAR$year
  }
  data <- HS_GENE_YEAR %>%
    filter(symbol %in% genes) %>%
    filter(year %in% years) %>%
    group_by(year) %>%
    count(symbol) %>%
    ungroup()
  ggplot(data) +
    geom_bar(aes(x=year, y=n, fill=symbol), 
             position = "dodge", stat = "identity")
  
}

绘制2000-2021 得基因符号词云图：

plotWC(keys = 2000:2021)

绘制TP53 发表年限词云图

plotWC(keys = c("TP53"), key="gene")

查看"APOE", “EGFR”, “ACE”, 2000-2021 发表文章数

genes <- c("APOE", "EGFR", "ACE")

plot_bar(genes, 2000:2021)