Repository navigation
Blast mapping file
Issue: 'Subject taxonomy id' and 'Taxonomy' not shown
(no changes made to Snakefile).
Probably caused by change in 'nucl_gb.accession2taxid' (taxdump.zip -> newtaxdump.zip).
Solution: the first column of the mapping_file (.map) should
be adjusted for each of the subsets.
Either by removing the version-number of the accession in the mapping file:
mv example.map example.map.back
sed 's/\.[0-9]\+ / /g' example.map.back > example.map
Or prefixing the first column of the mapping file with "lcl|"
(see prefixing ad hoc identifiers for releases prior to BLAST+ 2.9.0):
mv example.map example.map.back
awk '{print "lcl|"$0}' example.map.back > example.map
Issue: No mapping file created for bacterial_genome_selection
Solution: change "-m bacterial_accession" to "/NT/accession_taxonid"
Keep in mind that without modifying the Snakefile, makeblastdb has to be
run again (because no mapping file was created initially) and that the first
column of the mapping file has to be adjusted using one of the two methods
described above.
python create_mapping_file.py -f bacterial_genome_selection.fa
-m filepath/NT/accession_taxonid -o bacterial_accessions.map
Future prospects
Using Blastdb ver 5 allows blasting limited by taxonomy and/or targeted loci.
In theory this should make the use of subsets (bacterial_genome_selection,
12S, 16S) redundant. However not all of our subsets are currently availabe
within the targetedloci datasets (i.e. COI, matK ..)
Galaxy notes:
- 2to3
- Activate user
- Backup Repositories
- Blast dbs
- Blast db genbank
- Blast mapping
- Checkout release
- Conda envs
- Cron jobs
- Filemerge and opendiff
- Install Tools
- Galaxy training
- Git stuff
- Make Otu Table
- Merge Reads Flash
- Multi to Singleline fas
- Out of Space
- Restart Galaxy
- Rmate
- Set Locale
- Stdout
- Tool menu
- User statistics
- User statistics 2
- User statistics COMMpare
- Workflows