Elastic-blast

Improvements

- ncbi/elb docker image is now pulled from GCR and ECR repositories
- ElasticBLAST reports misspelled and misplaced config parameters
- Default number of CPUs is now 16 or the number of available CPUs
- Custom search tags are added to default ones
- Default disk size for AWS is now 1T

Bug fixes

- blast-tuner.py no longer reports aws-region for a GCP search
- blast-tuner.py recommends AWS instance types from M5, C5, and R5 families only

New features

- Upgrade to BLAST+ 2.12.0
- Experimental 2-stage cloud query split
- Record configuration in results bucket metadata
- Stress test added
- ElasticBLAST is available from PyPI and BioConda now

Improvements

- blast-tuner uses mem-limit and machine-type suggestions based on BLASTDB metadata
- OutOfMemoryError bug handled systematically

New features

- Implement tool to create BLASTDB metadata
- Distribute ElasticBLAST on PyPI.org: https://pypi.org/project/elastic-blast/0.1.4/

Improvements

- Add negative_taxidlist support to ElasticBLAST
- Use autoscaling by default for ElasticBLAST on GCP

Bug fixes

- Detect unbalanced quotation marks in BLAST options and report error

Improvements

- Added verbose option for elastic-blast status
- Selection of MT mode, number of CPUs and batch length in blast-tuner
- Elastic-blast can use instances with local SSD in AWS
- Elastic-blast works with "optimal" instance type in AWS
- Creation of PEX files on github
- Resubmission of an elastic-blast search elicits an error message

Bug fixes

- Cleaning of query batches on local machine's /tmp directory

Improvements

- Output file names unified over GCP and AWS versions

Bug fixes

- Retry logic implemented to handle database download errors and job structural timeouts