Supplementary datasets for PANNZER2 article

Test sequence sets

These represent the sequence sets that we used to test different scoring functions. We also used them to compare PANNZER2 and eggNOG-mapper. We represent separate set for testing each sub-ontology:
  • Biological Process
  • Molecular Function
  • Cellular Component

    Smaller sequence sets to test ARGOT2 web server:

  • Biological Process
  • Molecular Function
  • Cellular Component

    Evaluation: excluded GO classes

    This list of GO-classes were considered to be too large to represent informative annotation. Note that if a protein lacks a (specific) annotation the evaluation cannot tell correct and wrong annotations apart from each other.

    Results with PANNZER2

    These are the prediction tables obtained with PANNZER2, while using the sequence filtering. These correspond to above sequnce sets. Again each sub-ontology is represented separately:
  • Biological Process
  • Molecular Function
  • Cellular Component
    Following files represent the results without the sequence filtering:
  • Biological Process
  • Molecular Function
  • Cellular Component

    Results with eggNOG

    These are the prediction tables obtained with eggNOG-mapper. We used the sequence sets at beginning of this page. Again each sub-ontology is represented separately:
  • Biological Process
  • Molecular Function
  • Cellular Component

    Results with ARGOT2

    These are the prediction tables obtained with ARGOT2 web server. We used the smaller sequence sets at beginning of this page. Note that for method comparisons you have to filter the corresponding sequences from the larger eggNOG and PANNZER results. Again each sub-ontology is represented separately:
  • Biological Process
  • Molecular Function
  • Cellular Component

    PANNZER results with CAFA2 dataset

    Here we represent the predictions that PANNZER did on CAFA2 all.fasta dataset. CAFA supplementary material is here and CAFA matlab code here. Here we have only one set of predictions for PANNZER with filtering and one set of predictions for PANNZER without filtering. Lists, defined in matlab code, will select BP, MF or CC related sequences.