Reading a parquet file to an R data frame

R is a bit chaotic. Some users of the iq software reported issues when trying to use DIA-NN parquet files. I could not reproduce problem myself. Writing to a tsv file works for some users, but not all. It is frustrating.

It turns out that the arrow’s read_parquet function returns not only an R basic data frame, but also other data types used in the dplyr package. As a consequence, when dplyr is loaded, the behaviour of some base functions changes !

> is.numeric(df[, intensity_col])
[1] TRUE
> library(dplyr) 

Attaching package: ‘dplyr’

The following objects are masked from ‘package:stats’:

    filter, lag

The following objects are masked from ‘package:base’:

    intersect, setdiff, setequal, union
> is.numeric(df[, intensity_col])
[1] FALSE

Here df is read from a parquet file.

Long story short, we have updated the iq package to explicitly cast the input to a data frame. Hope it works!