Merge Intervals

setwd("C:/Users/shi-lab/Documents/SF3B1")
df = read.csv("CLL_FDR05_HL_column.csv", header = TRUE, sep = ,)

head(df)

##   Rank L_chr  L_start    L_end H_chr   H_start     H_end
## 1    1             NA       NA chr18   2667031   2688391
## 2    2             NA       NA chr18   2656261   2666868
## 3    3             NA       NA  chr5 130545857 130721235
## 4    4  chr6 90431190 90431245  chr6  90431191  90431244
## 5    5             NA       NA  chr2  33701775  33740076
## 6    6             NA       NA  chr4 152020867 152021578

df3 = data.frame(Rank=numeric(nrow(df)), chr=character(nrow(df)), start=numeric(nrow(df)), end=numeric(nrow(df)), stringsAsFactors = FALSE)

head(df)

##   Rank L_chr  L_start    L_end H_chr   H_start     H_end
## 1    1             NA       NA chr18   2667031   2688391
## 2    2             NA       NA chr18   2656261   2666868
## 3    3             NA       NA  chr5 130545857 130721235
## 4    4  chr6 90431190 90431245  chr6  90431191  90431244
## 5    5             NA       NA  chr2  33701775  33740076
## 6    6             NA       NA  chr4 152020867 152021578

df$L_chr = as.character(df$L_chr)
df$H_chr = as.character(df$H_chr)

head(df)

##   Rank L_chr  L_start    L_end H_chr   H_start     H_end
## 1    1             NA       NA chr18   2667031   2688391
## 2    2             NA       NA chr18   2656261   2666868
## 3    3             NA       NA  chr5 130545857 130721235
## 4    4  chr6 90431190 90431245  chr6  90431191  90431244
## 5    5             NA       NA  chr2  33701775  33740076
## 6    6             NA       NA  chr4 152020867 152021578

# df4 = data.frame(matrix(NA, ncol = 4, nrow = nrow(df)))
# head(df4)

for(i in 1:nrow(df)) {
  if(is.na(df[i,3])) {
      df3[i,1]=df[i,1]
      df3[i,2]=df[i,5]
      df3[i,3]=df[i,6]
      df3[i,4]=df[i,7]      
  }
    else {
      df3[i,1]=df[i,1]
      df3[i,2]=df[i,2]
      df3[i,3]=df[i,3]
      df3[i,4]=df[i,4]   
    }
}

head(df3, 10)

##    Rank   chr     start       end
## 1     1 chr18   2667031   2688391
## 2     2 chr18   2656261   2666868
## 3     3  chr5 130545857 130721235
## 4     4  chr6  90431190  90431245
## 5     5  chr2  33701775  33740076
## 6     6  chr4 152020867 152021578
## 7     7 chr19  35836030  35837070
## 8     8 chr17   6918476   6919813
## 9     9  chr1 145761177 145761178
## 10   10  chr1 145756613 145756614

# write.csv(df3, file="result_LH.csv")