Accessing an Endpoint Programmatically

This topic provides guidance on accessing Data on Demand endpoints programmatically by showing some example implementations using R and Python.

Authentication and Data Access

Connections to Data on Demand endpoints must be authenticated. Users can submit their Anzo username and password when accessing data. Ultimately the data that is available to users from OData endpoints is subject to the security and composition of the graphmart as configured in Anzo.

Accessing an Endpoint with R (Through RStudio)

The following example shows how to connect to an OData endpoint from RStudio. The example uses the R programming language to access a Data on Demand endpoint and pull in data via a standard dataframe. New or existing R scripts can then be used with the data.

The first step in accessing data from RStudio is to prepare the R script that will construct the target URL and retrieve the resulting information via HTTP. The example script below accesses a pre-configured "Sample Data" endpoint. The script has sections for filtering the results as well as expanding the selection to include information from multiple classes:

require("httr")
require("jsonlite")
require("rstudioapi")

user   <- rstudioapi::showPrompt("Username", "Enter Anzo username", "sysadmin")
pw     <- rstudioapi::askForPassword(paste("Enter password for",user,sep=" "))

## Data on Demand endpoint
odata  <- "https://10.100.0.10/dataondemand/Sample-Graphmart/Sample-Data"

## Start from Probe class
startClass  <- "Probe?"

## Filter results for Homo sapiens species
filterKw   <- "$filter="
filterVal  <- "Species eq 'Hs'"
urlify     <- URLencode(filterVal)
filterStr  <- paste(filterKw,urlify,sep="")

## Select properties of interest (FeatureID) from base class
selectKw   <- "&$select="
selectVal  <- "FeatureID"
selectStr  <- paste(selectKw,selectVal,sep="")

## Select properties of interest (symbol) from Gene class
## via corresponds_to property on base Probe class
expandKw   <- "&$expand="
expandClass  <- "corresponds_to"
expandProps  <- "symbol"
expSelStr   <- "$select="
expandStr   <- paste(expandKw,expandClass,"(",expSelStr,expandProps,")",sep="")

## Specify format
format  <- "&$format=json"

## Generate OData URL using fragments above
url <- paste(odata,startClass,filterStr,selectStr,expandStr,format,sep="")

## Access OData endpoint
resultRaw  <- GET(url, (authenticate(user,pw, type = "basic")))
resultTxt  <- content(resultRaw, "text")
resultJson  <- fromJSON(resultTxt, flatten = TRUE)

print(url)

## Read results into dataframe
resultDataFrame  <- as.data.frame(resultJson)
View(resultDataFrame)

Executing the above R script from RStudio results in a dataframe that represents columns from the Probe and Gene classes.

Accessing an Endpoint with Python (Through a Linux Terminal)

Many users have existing Python scripts to use with data in Anzo or a familiarity with Python that would make exploring, retrieving, and leveraging the data easier. The following example shows how to connect to an OData endpoint by executing a Python script from a Linux terminal.

The first step in accessing data using Python is to prepare the Python script that will construct the target URL and retrieve the resulting information via HTTP. The example script below accesses a pre-configured "Sample Data" endpoint. The script has sections for filtering the results as well as expanding the selection to include information from multiple classes (the same filter and class properties that were used in the R example above).

import requests
import getpass
from urllib.parse import urlparse

un = getpass.getpass(prompt='Username: ')
pw = getpass.getpass(prompt='Password: ')

## OData endpoint
# Data on Demand URL
odata = 'https://10.100.0.10/dataondemand/Sample-Graphmart/Sample-Data/'

## Start from Lease class
startClass = "Probe?"

## Filter results
filterKw  = "$filter="
filterVal = "Species eq 'Hs'"
urlify    = urlparse(filterVal)
filterStr = filterKw + urlify.geturl()

## Select properties of interest (start date, missed payments, lease status) from base class
selectKw  = "&$select="
selectVal = "FeatureID"
selectStr = selectKw + selectVal

## Select properties of interest (name, social security number, credit score) from Individual class
expandKw  = "&$expand="
expandClass = "corresponds_to"
expandProps = "symbol"
expSelStr   = "$select="
expandStr = expandKw + expandClass + "(" + expSelStr + expandProps + ")"

## Specify format
format = "&$format=text/csv"

## Generate OData URL using fragments above
url = odata + startClass + filterStr + selectStr + expandStr + format

## Access OData endpoint
r = requests.get(url, auth=(un, pw), verify=False)

print("URL")
print(url)
print("CONTENT")
print(r.content.decode('unicode_escape'))
print(type(r))
print(type(r.content))

In this example, the output is returned in CSV format (rather than JSON like the R example).

Related Topics