profile_comparison.py

import json
import base64
import contextlib

from flask import Blueprint, request, render_template, flash, Markup, url_for
from sqlalchemy.orm import noload

from conekt import cache
from conekt.forms.profile_comparison import ProfileComparisonForm
from conekt.helpers.chartjs import prepare_profiles, prepare_profiles_download
from conekt.models.expression.coexpression_clusters import CoexpressionCluster
from conekt.models.expression.profiles import ExpressionProfile
from conekt.models.relationships.sequence_cluster import SequenceCoexpressionClusterAssociation
from conekt.models.sequences import Sequence

profile_comparison = Blueprint('profile_comparison', __name__)


@profile_comparison.route('/cluster/<cluster_id>')
@profile_comparison.route('/cluster/<cluster_id>/<int:normalize>')
@cache.cached()
def profile_comparison_cluster(cluster_id, normalize=0):
    """
    This will get all the expression profiles for members of given cluster and plot them

    :param cluster_id: internal id of the cluster to visualize
    :param normalize: if the plot should be normalized (against max value of each series)
    """
    cluster = CoexpressionCluster.query.get(cluster_id)
    associations = SequenceCoexpressionClusterAssociation.query.\
        filter_by(coexpression_cluster_id=cluster_id).\
        options(noload(SequenceCoexpressionClusterAssociation.sequence)).\
        all()

    probes = [a.probe for a in associations]

    # get max 51 profiles, only show the first 50 (the extra one is fetched to throw the warning)
    profiles = ExpressionProfile.get_profiles(cluster.method.network_method.species_id, probes, limit=51)

    if len(profiles) > 50:
        flash(Markup(("To many profiles in this cluster only showing the <strong>first 50</strong>. <br />" +
                      "<strong>Note:</strong> The <a href='%s'>heatmap</a> can be used to with more genes and " +
                      "allows downloading the data for local analysis.") % url_for('heatmap.heatmap_cluster',
                                                                                  cluster_id=cluster_id))
              , 'warning')

    profile_chart = prepare_profiles(profiles[:50], True if normalize == 1 else False,
                                     ylabel='TPM' + (' (normalized)' if normalize == 1 else ''))

    # Get table in base64 format for download
    data = base64.encodebytes(prepare_profiles_download(profiles[:50],
                                                        True if normalize == 1 else False).encode('utf-8'))

    return render_template("expression_profile_comparison.html",
                           profiles=json.dumps(profile_chart),
                           normalize=normalize,
                           cluster=cluster,
                           data=data.decode('utf-8'))


@profile_comparison.route('/', methods=['GET', 'POST'])
def profile_comparison_main():
    """
    Profile comparison tool, accepts a species and a list of probes and plots the profiles for the selected
    """
    form = ProfileComparisonForm(request.form)
    form.populate_species()

    if request.method == 'POST':
        terms = request.form.get('probes').split()
        species_id = request.form.get('species_id')
        normalize = True if request.form.get('normalize') == 'y' else False

        probes = terms

        # also do search by gene ID
        sequences = Sequence.query.filter(Sequence.name.in_(terms)).all()

        for s in sequences:
            for ep in s.expression_profiles:
                probes.append(ep.probe)

        # make probe list unique
        probes = list(set(probes))

        # get max 51 profiles, only show the first 50 (the extra one is fetched to throw the warning)
        profiles = ExpressionProfile.get_profiles(species_id, probes, limit=51)

        not_found = [p.lower() for p in probes]
        for p in profiles:
            with contextlib.suppress(ValueError):
                not_found.remove(p.probe.lower())

            with contextlib.suppress(ValueError):
                not_found.remove(p.sequence.name.lower())

        if len(not_found) > 0:
            flash("Couldn't find profile for: %s" % ", ".join(not_found), "warning")

        if len(profiles) > 50:
            flash(Markup(("To many profiles in this cluster only showing the <strong>first 50</strong>. <br />" +
                          "<strong>Note:</strong> The <a href='%s'>heatmap</a> can be used to with more genes and " +
                          "allows downloading the data for local analysis.") % url_for('heatmap.heatmap_main')),
                  'warning')

        # Get json object for chart
        profile_chart = prepare_profiles(profiles[:50], normalize,
                                         ylabel='TPM' + (' (normalized)' if normalize == 1 else ''))

        # Get table in base64 format for download
        data = base64.encodebytes(prepare_profiles_download(profiles[:50], normalize).encode('utf-8'))

        return render_template("expression_profile_comparison.html",
                               profiles=json.dumps(profile_chart), form=form, data=data.decode('utf-8'))
    else:
        profiles = ExpressionProfile.query.filter(ExpressionProfile.sequence_id is not None).order_by(ExpressionProfile.species_id).limit(5).all()

        example = {
            'species_id': None,
            'probes': None
        }

        if len(profiles) > 0:
            example['species_id'] = profiles[0].species_id
            example['probes'] = ' '.join([p.sequence.name for p in profiles])

        return render_template("expression_profile_comparison.html", form=form, example=example)