Get phenotypes associated with a list of genes from high content screens¶

%pip install idr-py

import csv
import os
import pandas as pd
from tempfile import NamedTemporaryFile

import scipy
import numpy
from skimage import filters
import matplotlib.pyplot as plt
from idr import connection

import requests
import tifffile

INDEX_PAGE = "https://idr.openmicroscopy.org/webclient/?experimenter=-1"

# create http session
with requests.Session() as session:
    request = requests.Request('GET', INDEX_PAGE)
    prepped = session.prepare_request(request)
    response = session.send(prepped)
    if response.status_code != 200:
        response.raise_for_status()

URL = "https://idr.openmicroscopy.org/mapr/api/{key}/?value={value}&case_sensitive=false&orphaned=true"
SCREENS_PROJECTS_URL = "https://idr.openmicroscopy.org/mapr/api/{key}/?value={value}"
PLATES_URL = "https://idr.openmicroscopy.org/mapr/api/{key}/plates/?value={value}&id={screen_id}"
DATASETS_URL = "https://idr.openmicroscopy.org/mapr/api/{key}/datasets/?value={value}&id={project_id}"
IMAGES_URL = "https://idr.openmicroscopy.org/mapr/api/{key}/images/?value={value}&node={parent_type}&id={parent_id}"
ATTRIBUTES_URL = "https://idr.openmicroscopy.org/webclient/api/annotations/?type=map&image={image_id}"

TYPE = "gene"
KEYS = {"phenotype":
    ("Phenotype",
     "Phenotype Term Name",
     "Phenotype Term Accession",
     "Phenotype Term Accession URL")
}

def parse_annotation(writer, json_data, name, data_type):
    screen_name = "-"
    plate_name = "-"
    project_name = "-"
    dataset_name = "-"
    if data_type == 'datasets':
        project_name = name
    else:
        screen_name = name
     
    for p in json_data[data_type]:
        parent_id = p['id']
        if data_type == 'datasets':
            dataset_name = p['name']
        else:
            plate_name = p['name']
        qs3 = {'key': TYPE, 'value': gene,
                'parent_type': data_type[:-1], 'parent_id': parent_id}
        url3 = IMAGES_URL.format(**qs3)
        for i in session.get(url3).json()['images']:

            image_id = i['id']
            url4 = ATTRIBUTES_URL.format(**{'image_id': image_id})
            for a in session.get(url4).json()['annotations']:
                ontologies = []  # for ontology terms for a phenotype
                row = {}
                for v in a['values']:
                    if str(v[0]) in KEYS['phenotype']:
                        if str(v[0]) in ['Phenotype']:  # has phenotype
                            row[str(v[0])] = v[1]  # so create row

                        # if there are ontology mappings for the
                        # phenotype, add them to the ontologies list
                        ontList = ['Phenotype Term Name',
                                   'Phenotype Term Accession',
                                   'Phenotype Term Accession URL']

                        if str(v[0]) in ontList:
                            ontologies.extend([str(v[0]), str(v[1])])
                    if row:
                        if (len(ontologies) > 0):  # 1+ ontology mapping
                            row.update({'Gene': gene,
                                        'Screen': screen_name,
                                        'Plate': plate_name,
                                        'Image': image_id,
                                        'Project' : project_name,
                                        'Dataset': dataset_name})
                            # we have the start of a row now
                            # but we want to print out as many rows
                            # as there are ontology mappings
                            # so if there is mapping to 1 ontology term
                            # print 1 row, if there are 2 ontology terms
                            # print 2 rows etc
                            numberOfRows = len(ontologies)/6
                            # this is 3 pairs of ontology values per
                            # mapping, add the ontology mappings and print
                            n = 1
                            while (n <= numberOfRows):
                                row.update({ontologies[0]: ontologies[1],
                                            ontologies[2]: ontologies[3],
                                            ontologies[4]: ontologies[5]})
                                # remove that set of ontology mappings
                                ontologies = ontologies[6:]
                                writer.writerow(row)
                                n = n + 1

home = os.path.expanduser("~")
csvfile = NamedTemporaryFile("w", delete=False, newline='', dir=home, suffix=".csv")
try:
    fieldnames = [
        'Gene', 'Screen', 'Plate', 'Project', 'Dataset', 'Image',
        'Phenotype', 'Phenotype Term Name', 'Phenotype Term Accession',
        'Phenotype Term Accession URL']
    writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
    writer.writeheader()
    new_genes = []
    for g in genes:
        qs1 = {'key': TYPE, 'value': g}
        url1 = URL.format(**qs1)
        json = session.get(url1).json()
        for m in json['maps']: 
            new_genes.append(m['id'])
            
    for gene in new_genes:
        qs1 = {'key': TYPE, 'value': gene}
        url1 = URL.format(**qs1)
        json = session.get(url1).json()
        for m in json['maps']:
            qs2 = {'key': TYPE, 'value': gene, 'compound_id': m['id']}
            url2 = SCREENS_PROJECTS_URL.format(**qs2)
            json = session.get(url2).json()
            for s in json['screens']:
                gene = s['extra']['value']
                qs3 = {'key': TYPE, 'value': gene, 'screen_id': s['id']}
                url3 = PLATES_URL.format(**qs3)
                parse_annotation(writer, session.get(url3).json(), s['name'], 'plates')
            for p in json['projects']:
                gene = s['extra']['value']
                qs3 = {'key': TYPE, 'value': gene, 'project_id': p['id']}
                url3 = DATASETS_URL.format(**qs3)
                parse_annotation(writer, session.get(url3).json(), p['name'], 'datasets')           
finally:
    csvfile.close()

pixels = image.getPrimaryPixels()
image_plane = pixels.getPlane(0, 0, 0)

name="%s/%s.tif" % (home, image_id)
tifffile.imsave(name, predicted)

conn.close()