sequence-search

How to

Generate bowtie2 database

First download all sequences and associated occurrences from the OBIS database to data/sequences.csv and data/occurrences.csv:

create temp table sequence_hashes as
select
    dna.id,
    dna.occurrence_id,
    MD5(dna.flat->>'DNA_sequence') as hash,
    dna.flat->>'DNA_sequence' as sequence
from dna;

create temp table sequence_occurrences as
select
    sh.hash,
    round(occurrence.decimallongitude, 3) as decimalLongitude,
    round(occurrence.decimallatitude, 3) as decimalLatitude,
    occurrence.dataset_id,
    aphia.classification->>'phylum' as phylum,
    aphia.classification->>'class' as class,
    aphia.classification->>'order' as order,
    aphia.classification->>'family' as family,
    aphia.classification->>'genus' as genus,
    aphia.record->>'scientificName' as scientificName,
    count(*)
from sequence_hashes sh
left join occurrence on sh.occurrence_id = occurrence.id
left join aphia on occurrence.aphia = aphia.id 
group by
    sh.hash,
    round(occurrence.decimallongitude, 3),
    round(occurrence.decimallatitude, 3),
    occurrence.dataset_id,
    aphia.classification->>'phylum',
    aphia.classification->>'class',
    aphia.classification->>'order',
    aphia.classification->>'family',
    aphia.classification->>'genus',
    aphia.record->>'scientificName';

select * from sequence_occurrences; -- occurrences.csv

select distinct on (hash) hash, sequence -- sequences.csv
from sequence_hashes
order by hash;

Build the fasta file, occurrence sqlite database, and bowtie2 database:

./build_db.sh
rsync -r data ***@***:/data/sequence-search/data

Run the API

cd api
uvicorn main:app --reload --port 8086
# or
gunicorn --worker-class uvicorn.workers.UvicornWorker --bind '127.0.0.1:8086' --daemon main:app