- prokka/* - docker run --rm --user
$(id -u):$ (id -g) -v$PWD:/mnt staphb/prokka:latest prokka --outdir /mnt/prokka /mnt/sequences.fna - clinker/* - ./cluster.py | xargs -li docker run --rm --user
$(id -u):$ (id -g) -v $PWD:/mnt quay.io/biocontainers/clinker-py:0.0.23--pyh5e36f6f_0 bash -c 'clinker -j 10 -p /mnt/clinker/clusters{}' - ~/Downloads/clinker*.svg - Open dump_clinker.html in browser
- clinker_all.html - cat <(echo '<style>svg {clear: both;}</style>') ~/Downloads/clinker*.svg <(echo '') > clinker_all.html
Problem: Some strains have been sequenced multiple times, and therefore have multiple valid NCBI accessions associated with the same strain name
Solution: No longer deduplicate in cases where the same strain name has alternate accessions values associated with it - added as distinct entries in the Strain table if the accessions are unique.
Problem: Some strains do not have any accession value associated with them (this is often the case when the sequence for the GI/phage has its own accession and referencing the sequence of the strain is rendered unnecessary)
Solution: The database was altered slightly to allow for a null value associated with a strain. (Given the problem discussed above related to alternate accessions for a strain, it is possible to have multiple entries for strains with e same name, some with an accession/gbuid and one with a null value)
Problem: Some GIs are given uninformative names, often GI1,...,GI where GI is the nth GI discovered and characterized for that species. This can lead to ambiguous and confusing cases (eg. GI1 from Enterococcus faecium is not the same as GI1 from Burkholderia cenocepacia) and if these names are not distinguished in the output the user could reach inaccurate conclusions
Solution: Duplicate GIs in the database will have a name that combines their given name in the literature and a unique curated genomic island (cGI) ID assigned at the time of database development (ex. GI1_cGI_42).
Problem: Multiple publications were appended and separated by semicolons (usually) in the input data. These were being stored in a single field.
Solution: A separate publications table was generated to store the publication information so that each publication can have a unique entry. This is in keeping with standard database development practices.
Problem: Some GIs that were well studied and characterized were not assigned names by the publishing authors, but were still collected in the curation process.
Solution: These GIs will only have the ‘cGI#’ ID for their name
Problem: Each chromosome from a genome with multiple chromosomes will have a unique NCBI accession/entry associated with the same strain name. These were originally being deduplicated.
Solution: This problem will be solved by the implementation of the solution for the “Alternate Accessions for a Strain” solution, which now allows entries with the same strain name to be entered separately and keep their alternate accession IDs
Problem: In some cases, the accession associated with an entry points to a genome (from which the GI sequence spans a given set of coordinates) and in other cases the accession value points to an entry only containing the phage/GI sequence itself
Solution: In the case where the accession is associated with just the phage/GI sequence, the Sequence entry for that GI will have a gbuid that points directly to the NCBI entry (as opposed to the strain).
Problem: A record is missing a sequence fasta file
Solution: The GI accession or strain accession is searched and downloaded from the NCBI Nucleotide database. If the strain accession is used, the start and end coordinates must be present and only that coordinate range is retained in the fasta
Problem: Some records refer to the same GI
Solution: If records have the same GI accession, or strain accession with same start/end coordinates, then the GI will be deduplicated.
Problem: Some records have 100% sequence identity
Solution: Use name of first published after verifying gene content. Unless the later name is significantly more popular.
GIs with added tails are different GIs but should be clustered. Prefer longer alignments when searching database.
Colera toxin phage might be a good reference for examples for divergent naming and sequences and criteria for divergence.
GI names should be <28 characters to allow compatibility with various file formats