Custom classes to properly maintain metadata

[mahesh-panchal]

This is a suggestion that may not be feasible, but perhaps using custom objects might make handling meta data easier.

The issue is that meta map manipulation is a common feature of workflows because due to using the nf-core way of passing meta data around the workflow. This often means meta-maps are manipulated putting things in and out to get the correct fields to join on.

Being able to use custom classes might make the metadata handling situation better. As for the most part, these are simply data stores. The primary issue though is handling file staging. This maybe possible though by extending nextflow's ArrayBag class.

It's possible to use custom classes https://github.com/mahesh-panchal/nxf-custom-object-test at least as input and perform operations on.

Potential meta data objects

Sample:

• ID:
• Ploidy:
• Kmersize:
• taxon ID:

?

• Busco_lineage:

Read data:

• Read1
• Read2: null if no pair
• single_end: true if not paired end data.
• Kmercov
• readcov
• readgroup

Assembly:

• primary: hap1 or consensus sequence
• alternate

[mahesh-panchal]

Processes are not designed to take the output