benjiec / giraffe

Blast and Sequence Feature Detection

Giraffe consists of two tools: Giraffe feature detection, and Hippo feature database manager.

User can define features using Hippo, assign features to databases, build NCBI blast databases, and use Giraffe API to detect features in a given query sequence.

This software was originally written by Misha Wolfson and Benjie Chen, copyrighted by Addgene, and released under the MIT License. See LICENSE file. It is now maintained by Benjie Chen. Originally, Giraffe included a Javascript application for visualizing features on a sequence. That tool has now been moved to giraffe-ui.

Try it out

Requirements:

• System requirements: see provison/provision.sh (e.g. you can use this to provision a Vagrant instance).

• Python requirements: pip install -r requirements.txt (Currently supports Django 1.5 or higher)

• NCBI: install NCBI Blast toolkit (this takes awhile): cd ncbi; . install

Install Django service and build default database:

git clone [email protected]:benjiec/giraffe.git
cd giraffe
(cd src; python manage.py migrate)
(cd src; python manage.py build_blastdb)

Run test server:

cd src; python manage.py runserver 0.0.0.0:8000

To test it out, install giraffe-ui and point the example analyze.html and draw.html files to

http://0.0.0.0:8000/giraffe/analyze/

API

You can POST a sequence to "/giraffe/analyze/", with the following params:

db=default&sequence=your_dna_sequence_or_genbank_sequence_here

There are some other params you can set, e.g. identity_threshold, feature_threshold, circular. See giraffe/views.py and giraffe/features.py.

This will return a JSON array of [sequence_len, array of features, sequence]. Each feature in the feature array has, among other attributes

{
  "query_start": ...,
  "query_end": ...,
  "subject_start": ...,
  "subect_end": ...,
  "label": ...,
  ...
}

Features from blast also has the following attributes

{
  ...
  "alignment": {
    "query": ...,
    "match": ...,
    "subject": ...
  }
}

The format of the JSON is "standardized" in giraffe-features.

By default, feature detection returns features (from blast), and for DNA sequences, restriction sites and ORFs. If you set blastonly=1 in the URL, it will not detect restriction sites or ORFs. You can also set input=protein and send in a protein sequence as query.

You can use Hippo to build DNA or protein blast databases. Giraffe uses blastn to blast DNA queries against DNA blast databases, blastx to blast DNA queries against protein blast databases, blastp to blast protein queries against protein blast databases, tblastn to blast protein queries against DNA blast databases.

Programmatically, you can construct new blast databases using Hippo. E.g.

from hippo.models import Feature, Feature_Type, Feature_Database
db = Feature_Database(name='my_db')
db.save()
ft = Feature_Type.objects.get(type='Promoter')
feature = Feature(type=ft, name='Promoter1', sequence='AGCTATTTCGGAGTCGGCGATTACGATCGGCGATCG')
feature.save()
db.features.add(feature)
db.build()

You can then query against database. E.g.

from hippo.models import Feature_Database
from giraffe.features import blast

db = Feature_Database.objects.get(name='my_db')
query = '...' # some query sequence
features = blast(query, db)

Hippo as a BLAST frontend

Use Django admin to add features, group features into databases. Use the build_blastdb django management command to build your databases, then Giraffe can use the databases for feature detection. Alternatively, call the build method on a Feature_Database object directly.