kdm9 / onlign

Online alignment prototypes for ANU improvements to AUGUR

onlign

Online alignment prototypes for ANU improvements to AUGUR

Install deps

conda env create -f environment.yml && conda activate onlign

run GISAID ncov

mkdir data/
wget -O data/gisaid_cov2020_sequences.fasta  $GISAID_DATA_URL

# see `bash ./alignment.sh` for advanced options
bash alignment.sh data/gisaid_cov2020_sequences.fasta 

TODOs

• A more robust way of detecting the N most diverse samples that doesn't pick long tips or otherwise strange sequences

  • By which I mean prefiltering the alignments somehow so that the guide tree doesn't include strange samples

• remove known-dodgy sites and samples from alignment

• smarter handling of alignment funkyness that maintain compatibility with the recognised coordinate space

  • Alignment funkyness e.g. regions gap-or-n-only columns due to funky samples

• Verify that the "core" alignment matrix doesn't change between new sequences before just concatenating the new seqs together (in gatherprofilealn.py)

• Integrate treebuilding logic a la Rob's state machine diagram

• run with bits of Sebastian's 100k seq simulation