ideas - future improvments

[d-courtine]

Here is a non-exhaustive list of idea I could implement:

• TransDecoder-Predict capture the log and check if it ends correctly
  otherwise, re-run it with the parameter --no_refine_starts.
• Clarify the 3 steps: use the right vocabulary, eg "peptide sequence"
  instead of "cds"
• add a requirements.txt ?
• check if a steps ended properly:
  • trimmomatic output the information in STDOUT/STDERR
    (TrimmomaticPE: Completed successfully)
  • Trinity too ("All commands completed successfully. :-)")
• Tweak tool parameters by passing them to the command line:
  • trimmomatic MINLEN:32 SLIDINGWINDOW:10:20 LEADING:5 TRAILING:5
  • MMseqs2: cluster mode; sensitivity, min seq ID, ...
• Add a dict to store the subpath values
  • eg:
    • {"00" : "00_FastQC_raw",
    • "01" : "01_trimmomatic",
    • ... }
• Add Tiara for the identification of
  Eukaryotic data : the aim would be to drop prokaryotic sequences
• clean krypton/tasks/ko_annot.py and make it usable alone
  • Add arguments parser for this script!