- fix fail of analysis if no sample groups were specified

- fix parsing of Proteome discoverer output that lacks a column named "Master" by calculating the mean value of the non-zero protein abundancies for all protein groups

- fix dbcan tasks: erroneous parsing of hmmer output resulted in many missed annotations. Previously the wrong columns (1/3) were parsed from hmmer results, which resulted in a lot of missed hits. The correct columns are 0 and 2 for hmmscan and hmmsearch respectively.
- internal: refactor building and writing of summary for comprehensibility.
- internal: remove second accession column from id2annot map files
- internal: rename column names of dbcan, pfam, tigrfam maps

- fix tigrfam download links
- fix memory issues during lca calculation. Large numbers of protein groups in combination with many samples lead to large RAM consumption. Protein Group and Sample information is now stored in a sqlite db (`pgs/protein_groups_db.sql`), reducing the memory requirements to a minimum.
- fix order of summary columns: When using sample groups, the name and title of summary columns did not match (affected: quant and mafft columns). Krona plots were not affected.
- fix Proteome Discoverer Parsing for groups without master protein (#84). If a group has no master protein, it's abundance is set to the mean of all member protein abundance values
- fix emapper v4 analysis (set block size parameter only for emapper v5)

- fix #82: prophane.de: eggnog jobs fail with memory error for some query fasta files
  - fixed by setting the block_size parameter based on the size of the all.faa fasta file: if the fasta file is larger than 10MB: block_size = 10 / size_in_MB(all.faa), rounded to one decimal. For smaller fasta files, the block_size is set to 2.

- fix eggnog v5 download
- fix eggnog v5 result mapping
- fix emapper log redirection, now is properly written to {task_file_name}.log

- add support for eggnog database version 5.0.2

- fix issue 80: only download ncbi_taxdump database for taxonomic analyses
- fix mafft workflow if no protein groups with more than one accession are present
- fix prophane crash upon executing `prophane list-dbs` on outdated databases. DBs are now automatically migrated.
- add `prophane --version` parameter to cli
- doc: adapt installation instructions to include direct conda installation, remove setup.sh

- add support for gzipped fasta files as input
- fix crash during parsing of Proteome Discoverer Output if it contains protein groups without any quantification values for the master protein. Now, Prophane ignores these protein groups.

add support for large mzIdentML files

refactor code to parse search result using snakemake (previously: plain python)

- fix command line interface option "prepare-dbs". It now accepts additional snakemake options and will work if the job config contains acc2annot_mapper tasks

- fix database migration for setups that contain multiple versions of the same db

- fix prophane path detection in CLI

** 6.0
*** Breaking Changes
change of command line interface:

    prophane -> prophane.py run
    prophane --list-dbs -> prophane.py list-dbs
    prophane --list-styles -> prophane.py list-styles

*** Features
Automatically download databases that are specified in the job-config. On first run of prophane, run prophane init {DB_DIR} where DB_DIR is an empty or non existant directory. To execute prophane, run prophane run {path-to-job-config}.

*** Changes
* bump the db_schema version to 5
* task and plot files now include the database version number instead of the md5sum: tasks/{annot_type}_annot_by_{tool}_on_{db_type}.v{db_version}.task{taskid}.{ext}

- add mamba dependency

- add parser for proteome discoverer output

- fix mztab parser skipping samples without associated spectra and not counting all spectra

fix parsing of spectra IDs from protein group yaml and some test files

LCA determination of proteins with multiple annotations: If a species/lineage is found for all protein accessions in a protein group, this species/lineage is chosen as LCA. Before: if any other species was determined --> various.

LCA determination with two different methods:
1. (default) per protein group (with threshold, default: 1). If the ratio of proteins assigned to an ancestor is at least as high as the threshold, this ancestor is assigned to the entire group (multiple above threshold --> lca with highest support). Changes previous default behaviour. For example if a species/lineage is found for all protein accessions in one group, this species/lineage has a support of 100% and is chosen by the new lca-methode as LCA. Before: if any other species was determined --> various.
2. democratic: takes the one with highest occurrence among all protein groups in the whole task

New columns in summary.txt: lca-support, describing the number of proteins/spectra assigned to the respective LCA. Also the summary rule is adjusted for csv export, we only have to adjust the separator and includes spectra in output (if available).