如何提高比较两个列表和一个范围之间的值的python脚本的速度?

时间:2017-03-16 23:16:53

标签: python

我有两个大型文件数据集:

File1:
Gen1 1 1 10
Gen2 1 2 20
Gen3 2 30 40

File2:
A 1 4
B 1 15
C 2 2

预期产出:

Out:
Gen1 1 1 10 A 1 4
Gen2 1 2 20 B 1 15

现在我有代码,基本上只是试图找到文件2在文件1中的实例,如果file2 [1]匹配file1 [1]并且落在文件1中的范围之间。

我的代码如下:

for i in file1:

    temp = i.split()

    for a in file2:

        temp2 = a.split()

        if temp[1] == temp2[1] and temp2[2] >= temp[2] and temp2[2] <= temp[3]

             print(i + " " + a + "\n")

        else:

            continue

代码有效,但我觉得它需要的时间比它应该长很多。有没有更简单的方法或方法来做到这一点?我觉得有一些巧妙地使用我不做的地图或哈希。

谢谢!

1 个答案:

答案 0 :(得分:0)

熊猫可能是个不错的选择。请参阅this示例。

当文件很大时,我更喜欢sqlite而不是pandas。可以从sqlite DB加载Pandas数据帧。

import sqlite3

file1 = """Gen1 1 1 10
Gen2 1 2 20
Gen3 2 30 40"""

file2 = """A 1 4
B 1 15
C 2 2"""

# your code (fixed)
print("desired output")
for i in file1.splitlines():
    temp = i.split()
    for a in file2.splitlines():
        temp2 = a.split()
        if temp[1] == temp2[1] and int(temp2[2]) >= int(temp[2]) and int(temp2[2]) <= int(temp[3]):
            print(i + " " + a)


# Make an in-memory db
# Set a filename if your files are too big or if you want to reuse this db
con = sqlite3.connect(":memory:")
c = con.cursor()

c.execute("""CREATE TABLE file1
(
    gene_name text,
    a integer,
    b1 integer,
    b2 integer
)""")

for row in file1.splitlines():
    if row:
        c.execute("INSERT INTO file1 (gene_name, a, b1, b2) VALUES (?,?,?,?)", tuple(row.split()))

c.execute("""CREATE TABLE file2
(
    name text,
    a integer,
    b integer
)""")

for row in file2.splitlines():
    if row:
        c.execute("INSERT INTO file2 (name, a, b) VALUES (?,?,?)", tuple(row.split()))

# join tow tables
print("sqlite3 output")
for row in c.execute("""SELECT
    file1.gene_name,
    file1.a,
    file1.b1,
    file1.b2,
    file2.name,
    file2.a,
    file2.b
FROM file1
JOIN file2 ON file1.a = file2.a AND file2.b >= file1.b1 AND file2.b <= file1.b2
"""):
    print(row)

con.close()

输出:

desired output
Gen1 1 1 10 A 1 4
Gen2 1 2 20 A 1 4
Gen2 1 2 20 B 1 15
sqlite3 output
(u'Gen1', 1, 1, 10, u'A', 1, 4)
(u'Gen2', 1, 2, 20, u'A', 1, 4)
(u'Gen2', 1, 2, 20, u'B', 1, 15)