CODEWITHSUNDEEP
rplyr
Davy Kavanagh
Davy Kavanagh

Reputation: 4939

Selecting the min of a group with ddply

My data frame looks like this

 TFexp
             ID Entrez_Gene_ID Symbol  adj.P.Val     P.Value          t         B       logFC GSM507414
  ILMN_1765574           7020 TFAP2A 0.03071446 0.001797766 -5.9919321 -1.012015 -1.02847949  9.537218
  ILMN_2374115           7020 TFAP2A 0.03914067 0.002813079 -5.4197845 -1.517871 -0.75604907  6.776104
  ILMN_1749838           7593  ZNF42 0.12390434 0.021504872 -3.2861751 -3.807814 -0.48844495  8.586840
  ILMN_2181540           7528    YY1 0.21007443 0.054908628  2.4869982 -4.834348  0.31190866 10.404929
  ILMN_1733135           7020 TFAP2A 0.22665655 0.063170290 -2.3738597 -4.984464 -0.82238355  3.827819
  ILMN_1670816           7593  ZNF42 0.44843565 0.220579380  1.3975294 -6.251177  1.60730795  2.963474
  ILMN_1770892           7528    YY1 0.47245442 0.243567365  1.3198784 -6.343075  0.16865034 10.903882
  ILMN_2275760           7020 TFAP2A 0.72242186 0.532549376 -0.6695509 -6.978362 -0.52387868  3.392317
  ILMN_1676010           6667    SP1 0.75701851 0.581434509  0.5886784 -7.034562  0.07281042  7.846117
  ILMN_2282477           7020 TFAP2A 0.95496669 0.907458427 -0.1221984 -7.225401 -0.02493289  6.903279

Within each level for Symbol I want to select the min, and construct a new data frame from these minimums. I tried this

ddply(TFexp, .(Symbol), function(x)x[x$P.Value==min(x$P.Value)])

but this returned the following:

             ID Entrez_Gene_ID Symbol adj.P.Val    P.Value          t         B      logFC GSM507414 GSM507415
1  ILMN_1676010           6667    SP1 0.7570185 0.58143451  0.5886784 -7.034562 0.07281042  7.846117  7.920055
2  ILMN_1765574           <NA>   <NA>        NA         NA -5.9919321        NA         NA        NA        NA
3  ILMN_2374115           <NA>   <NA>        NA         NA -5.4197845        NA         NA        NA        NA
4  ILMN_1733135           <NA>   <NA>        NA         NA -2.3738597        NA         NA        NA        NA
5  ILMN_2275760           <NA>   <NA>        NA         NA -0.6695509        NA         NA        NA        NA
6  ILMN_2282477           <NA>   <NA>        NA         NA -0.1221984        NA         NA        NA        NA
7  ILMN_2181540           <NA>    YY1        NA 0.05490863         NA -4.834348         NA 10.404929        NA
8  ILMN_1770892           <NA>    YY1        NA 0.24356737         NA -6.343075         NA 10.903882        NA
9  ILMN_1749838           <NA>  ZNF42        NA 0.02150487         NA -3.807814         NA  8.586840        NA
10 ILMN_1670816           <NA>  ZNF42        NA 0.22057938         NA -6.251177         NA  2.963474        NA

Any ideas?
Cheers,
Davy.

Upvotes: 2

Views: 1846

Answers (2)

Joshua Ulrich
Joshua Ulrich

Reputation: 176638

That's because x[x$P.Value==min(x$P.Value)] is selecting a column of the data.frame, not a row. Use x[x$P.Value==min(x$P.Value),] instead. Better yet, use x[which.min(x$P.Value),].

Upvotes: 2

Brian Diggs
Brian Diggs

Reputation: 58825

(untested)

ddply(TFexp, .(Symbol), function(x)x[which.min(x$P.Value),])

The two changes:

  1. added a , after the row selection criteria to make it actually be a row selection.
  2. changed the selection from a logical comparison to avoid the possibility of problems with comparisons of floating point numbers.

Upvotes: 3

Related Questions