SQL WHERE NOT EXISTS查询不返回结果

时间:2015-04-09 17:11:24

标签: mysql sql sql-server impala

我无法构建查询并让它在impala上运行。我创建了以下工作查询来连接两个表:

SELECT *
FROM illuminavariant as vcf, ensembl_genes as ens 
WHERE vcf.filter = "PASS" 
AND vcf.qual > 100 
AND vcf.chromosome = ens.chromosome
AND vcf.position BETWEEN ens.start AND ens.stop

现在我正在尝试编写一个查找所有变体的查询WHERE vcf.filter =“PASS”和vcf.qual> 100,但没有染色体和位置的匹配。

我试过这个:

SELECT * 
FROM p7dev.illumina_test, p7dev.ensembl_test
WHERE NOT EXISTS(
  SELECT * 
  FROM p7dev.illumina_test as vcf, p7dev.ensembl_test as ens
  WHERE vcf.chromosome = ens.chromosome  
  AND vcf.position BETWEEN ens.start AND ens.stop  
 )

但这并没有带来任何结果。我认为一个WITH子句可能会有所作为,但我真的很感激,如果有人能帮助我理解这将如何工作的逻辑。非常感谢!

2 个答案:

答案 0 :(得分:2)

由于您正在寻找与任何集合无关的变体,因此您可以形成变体和集合的交叉连接来过滤掉行,这似乎很奇怪。如果这真的是你想要的,那么应该这样做:

SELECT *
FROM illuminavariant as vcf, ensembl_genes as ens 
WHERE vcf.filter = "PASS" 
AND vcf.qual > 100 
AND (
    vcf.chromosome != ens.chromosome
    OR vcf.position < ens.start
    OR vcf.position > ens.stop
  )

这只是否定了将变量行与整体行相关联的条件。

我怀疑你真正想要的更像是这样:

SELECT vcf.*
FROM
  illuminavariant as vcf
  LEFT JOIN ensembl_genes as ens 
    ON vcf.chromosome = ens.chromosome
    AND vcf.position BETWEEN ens.start AND ens.stop
WHERE
  vcf.filter = "PASS" 
  AND vcf.qual > 100
  AND ens.chromosome IS NULL

执行与第一个查询相同的连接,但作为左连接。实际表示匹配的行然后由ens.chromosome IS NULL条件过滤掉。它只返回变量表的列,因为整个点是找到在集合表中没有对应行的变体。

答案 1 :(得分:1)

试试这个......

SELECT * 
FROM p7dev.illumina_test vcf
WHERE NOT EXISTS( SELECT 1
                  FROM p7dev.ensembl_test as ens
                  WHERE vcf.chromosome = ens.chromosome  
                  AND vcf.position BETWEEN ens.start AND ens.stop
                 )
AND vcf.filter = 'PASS' 
AND vcf.qual > 100