This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Strange esearch behavior

This is a follow-up (kinda) to my previous question: Exclude Replaced or Discontinued UIDs from an reutils.esearch operation

I have a function where I

  • Call esearch on a query
  • Check if it has $no_errors() and then check if only one result is returned (length(uid(esearch())) == 1)
  • If the above conditions are satisfied, perform an esummary() call.

There's this search that throws off my algorithm: FAM231D [Gene Name] AND Homo sapiens[ORGN]. It returns 1 result, where the UID is NA. I feel cheated, like I got a gift-wrapped empty box in the mail. It's either supposed to error out if no results are found, or at least give me length(uid()) = 0.

Code:

> esearch('FAM231D', db='gene')
Object of class ‘esearch’
List of UIDs from the ‘gene’ database.
 [1] "103091866" "102724235" "100996721" "746217   " "113220841" "111543928"
 [7] "111535250" "108538690" "107000807" "105730623" "105709924" "105707698"
[13] "105707697" "105597670" "105549803" "105532204" "105478985" "104667973"
[19] "103889731" "102145358" "101147160" "100996741" "100986275" "100591271"

> esearch('FAM231D AND Homo sapiens[ORGN]', db='gene')
Object of class ‘esearch’
List of UIDs from the ‘gene’ database.
[1] "103091866" "102724235" "100996721" "100996741"

> esearch('FAM231D [Gene Name] AND Homo sapiens[ORGN]', db='gene')
Object of class ‘esearch’
List of UIDs from the ‘gene’ database.
[1] "NA"

> esearch('FAM231D [Gene Name] AND Homo sapiens[ORGN]', db='gene')$no_errors()
[1] TRUE

> uid(esearch('FAM231D [Gene Name] AND Homo sapiens[ORGN]', db='gene'))
[1] NA

> length(uid(esearch('FAM231D [Gene Name] AND Homo sapiens[ORGN]', db='gene')))
[1] 1

What's going on here?

ncbi eutils esearch
• 1,940 views • link
written by Ram Ram

But there are no results for your query:

> esearch('FAM231D[Gene Name] AND Homo sapiens[ORGN]', db='gene', rettype = "count")
Object of class ‘esearch’ 
[1] 0
• 1 views • link
written by vkkodali_ncbi vkkodali_ncbi

Sure, rettype=count gives us 0, but then why does rettype=uilist have a UI entry? I don't wish to perform multiple esearch() calls per keyword, so should the search not be consistent regardless of the rettype?

For example, try esearch('XXYYZZ [Gene Name] AND Homo sapiens[ORGN]', db='gene', rettype='uilist')

• 1 views • link
written by Ram Ram

1 answer

I cannot say what the motivation was for this behavior in reutils. From the linux command line, both of the following queries work in the same way: 

$ esearch -db gene -q 'FAM231D[Gene Name] AND Homo sapiens[ORGN]'
<ENTREZ_DIRECT>
  <Db>gene</Db>
  <WebEnv>NCID_1_47230685_130.14.18.34_9001_1540567994_1662557237_0MetA0_S_MegaStore</WebEnv>
  <QueryKey>1</QueryKey>
  <Count>0</Count>
  <Step>1</Step>
</ENTREZ_DIRECT>
$ esearch -db gene -q 'XXYYZZ[Gene Name] AND Homo sapiens[ORGN]'
<ENTREZ_DIRECT>
  <Db>gene</Db>
  <WebEnv>NCID_1_149285401_130.14.18.125_9001_1540578340_428880606_0MetA0_S_MegaStore</WebEnv>
  <QueryKey>1</QueryKey>
  <Count>0</Count>
  <Step>1</Step>
</ENTREZ_DIRECT>

That said, you can probably check for uid != "NA" in addition to length(uid(esearch())) == 1 to get around this issue.

• 1 views • link
written by vkkodali_ncbi vkkodali_ncbi

I guess I'll have to use the work-around for now. Thank you! If you could move your comment to an answer (using the moderate option), I'll accept it.

• 1 views • link
written by Ram Ram

Log in to answer this question.

More posts like this