Parquet needs some standarization

[ypriverol]

We need to do some standardization for the Parquet format that enables other people to understand the file format.

[bgruening]

Yeah, that would be nice and give it a proper name :)

[sorenwacker]

I like 'parquet', as it is pretty clear what library to use to open it.

Regarding column names. I had a few thoughts:

• The column name Mass or Masses is technically wrong as it is M/Z values. Or do you convert the M/Z values into masses internally?

• Intensities could be Intensity even if it is an array.

• RetentionTime was used in mzXML files, in mzML files I have seen it as ScanTime which is a bit more general and may be more accurate. It would not imply that a chromatographic step was used.

• Things like TIC are maybe convenient, but also somewhat redundant and it could be calculated easily in one line of code if the data would be in long format.

  df_long.groupby('scan_time_min').sum().plot(y='intensity')

I am quite new to metabolomics/proteomics thought. I am looking at the problem more from a data science Python-biased perspective.