按第一列定义的间隔有效地平均第二列

Question

数据文件中有两个数字列.我需要按第一列的间隔(例如100)计算第二列的平均值.

我可以用R编程这个任务,但我的R代码对于一个相对较大的数据文件来说真的很慢(数百万行,第一列的值在1到33132539之间变化).

在这里,我展示了我的R代码.我怎么能把它调到更快？其他解决方案是perl,python,awk或shell.

提前致谢.

(1)我的数据文件(制表符分隔,数百万行)

5380    30.07383\n
5390    30.87\n
5393    0.07383\n
5404    6\n
5428    30.07383\n
5437    1\n
5440    9\n
5443    30.07383\n
5459    6\n
5463    30.07383\n
5480    7\n
5521    30.07383\n
5538    0\n
5584    20\n
5673    30.07383\n
5720    30.07383\n
5841    3\n
5880    30.07383\n
5913    4\n
5958    30.07383\n

(2)我想得到的,这里间隔= 100

intervals_of_first_columns, average_of_2nd column_by_the_interval
100, 0\n
200, 0\n
300, 20.34074\n
400, 14.90325\n
.....

(3)R代码

chr1 <- 33132539 # set the limit for the interval
window <- 100 # set the size of interval

spe <- read.table("my_data_file", header=F) # read my data in
names(spe) <- c("pos", "rho") # name my data 

interval.chr1 <- data.frame(pos=seq(0, chr1, window)) # setup intervals
meanrho.chr1 <- NULL # object for the mean I want to get

# real calculation, really slow on my own data.
for(i in 1:nrow(interval.chr1)){
  count.sub<-subset(spe, chrom==1 & pos>=interval.chr1$pos[i] & pos<=interval.chr1$pos[i+1])
  meanrho.chr1[i]<-mean(count.sub$rho)
}

Answer 1

您根本不需要设置输出data.frame,但如果需要,可以.这是我将如何编码它,我保证它会很快.

> dat$incrmt <- dat$V1 %/% 100
> dat
     V1       V2 incrmt
1  5380 30.07383     53
2  5390 30.87000     53
3  5393  0.07383     53
4  5404  6.00000     54
5  5428 30.07383     54
6  5437  1.00000     54
7  5440  9.00000     54
8  5443 30.07383     54
9  5459  6.00000     54
10 5463 30.07383     54
11 5480  7.00000     54
12 5521 30.07383     55
13 5538  0.00000     55
14 5584 20.00000     55
15 5673 30.07383     56
16 5720 30.07383     57
17 5841  3.00000     58
18 5880 30.07383     58
19 5913  4.00000     59
20 5958 30.07383     59

> with(dat, tapply(V2, incrmt, mean, na.rm=TRUE))
      53       54       55       56       57       58       59 
20.33922 14.90269 16.69128 30.07383 30.07383 16.53692 17.03692 

您可以完成更少的设置(使用以下代码跳过incrmt变量:

    > with(dat, tapply(V2, V1 %/% 100, mean, na.rm=TRUE))
      53       54       55       56       57       58       59 
20.33922 14.90269 16.69128 30.07383 30.07383 16.53692 17.03692 

如果您希望结果可用于某些内容:

by100MeanV2 <- with(dat, tapply(V2, V1 %/% 100, mean, na.rm=TRUE))