Mercurial > repos > recetox > matchms_formatter
diff formatter.py @ 3:574c6331e9db draft
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 38a5028a7abe99794086e9b1374ab4bb8bfa68de
| author | recetox |
|---|---|
| date | Wed, 21 Sep 2022 15:29:51 +0000 |
| parents | 60f34912b3de |
| children | 966b4134ad12 |
line wrap: on
line diff
--- a/formatter.py Thu Sep 08 14:49:19 2022 +0000 +++ b/formatter.py Wed Sep 21 15:29:51 2022 +0000 @@ -1,5 +1,5 @@ import click -from pandas import DataFrame, read_csv +from pandas import DataFrame, read_csv, to_numeric def create_long_table(data: DataFrame, value_id: str) -> DataFrame: @@ -73,8 +73,8 @@ Returns: DataFrame: Joined dataframe on compounds containing scores an matches in long format. """ - matches = read_csv(matches_filename, sep=None, index_col=0) - scores = read_csv(scores_filename, sep=None, index_col=0) + matches = read_csv(matches_filename, sep="\t", index_col=0, header=0).apply(to_numeric) + scores = read_csv(scores_filename, sep="\t", index_col=0, header=0).apply(to_numeric) scores_long = create_long_table(scores, 'score') matches_long = create_long_table(matches, 'matches') @@ -113,11 +113,8 @@ @cli.resultcallback() def write_output(result: DataFrame, scores_filename, matches_filename, output_filename): - input_file = read_csv(scores_filename, sep=None, iterator=True) - sep = input_file._engine.data.dialect.delimiter - result = result.reset_index().rename(columns={'level_0': 'query', 'compound': 'reference'}) - result.to_csv(output_filename, sep=sep, index=False) + result.to_csv(output_filename, sep="\t", index=False) if __name__ == '__main__':