Processing with multiple column of features.tsv.gz

[wook2014]

I have recently deal with a perturb-seq data from 10X CRISPR-Screening and output from cellranger. the features file have 3 columns: ensemble, gene_name, feature_type. When I read in the by Read10X, it said:
10X data contains more than one type and is being returned as a list containing matrices of each type.
and therefore I meet up with a problem:

Error in cell_KO_hash[[colnames(perturb_seq)[i]]] : 
  wrong arguments for subsetting an environment

it seemed that the features file can only have 2 columns. can this problem be avoided for reducing doubts？