All of our input commands require CSV files, but we do not provide clear error messages or validation when users provide a TSV file instead. This specific issue happened after I tried to run pathogen-embed on a distance matrix produced by snp-dists which is a TSV file. Instead of providing an error when the input only had one column or a filename extension of ".tsv", pathogen-embed provided this obscure error message from sklearn's internals: ValueError: at least one array or dtype is required
We should at least check the filename extension or even use the built-in CSV sniffer to check whether the input files are formatted properly.
All of our input commands require CSV files, but we do not provide clear error messages or validation when users provide a TSV file instead. This specific issue happened after I tried to run
pathogen-embedon a distance matrix produced by snp-dists which is a TSV file. Instead of providing an error when the input only had one column or a filename extension of ".tsv", pathogen-embed provided this obscure error message from sklearn's internals:ValueError: at least one array or dtype is requiredWe should at least check the filename extension or even use the built-in CSV sniffer to check whether the input files are formatted properly.