cs_formats() is the machine-readable index of supported
reader grammars. Each adapter preserves the source column name in
dictionary$source_name and records its adapter and version
in provenance.
cs_formats()
#> format adapter_version
#> 1 table 1.0.0
#> 2 qupath 1.0.1
#> 3 qupath_tiled 1.0.1
#> 4 mcquant 1.0.0
#> 5 segmantr 1.0.1
#> 6 inform 1.0.0
#> description
#> 1 Generic delimited table with an explicit column map.
#> 2 QuPath colon-delimited measurement export
#> 3 Tiled QuPath tables with centroid ownership
#> 4 MCMICRO MCQuant measurement export
#> 5 segmantR feature table
#> 6 inForm cell segmentation data
#> tested_with status
#> 1 CSV, TSV and gzip-compressed delimited files stable
#> 2 QuPath 0.6.0 and 0.7.0 experimental
#> 3 QuPath 0.7.0 experimental
#> 4 MCMICRO/MCQuant experimental
#> 5 segmantR 0.2.0 experimental
#> 6 inForm/phenoptr experimentalThe QuPath adapter accepts both channel-first headers such as
DAPI: Mean and compartment-first headers such as
Nucleus: CD3e: Mean. The tiled adapter adds an ownership
pass over outer tile boxes; halo rows are counted and dropped. MCQuant
and segmantR readers treat centroid_col as x and
centroid_row as y, and require pixel calibration.
When no grammar is decisive, cs_read(format = "auto")
returns a classified ambiguity error and lists the candidates. This
prevents a plausible but wrong column match from reaching downstream
analysis.