Franklin API

This FastAPI API provides the answers to problems in the Rosalind problem set.

A deployed version of this API can be found here hosted on Azure. You can view the docs and test out some of the endpoints there.

Don't cheat!

I set this up as a personal project to explore FastAPI and solve some of the Rosalind problems. While you could use this to cheat on the Rosalind problems, I urge you not to. The whole point of the Rosalind problem set is to learn the algorithms and concepts behind some bioinformatics problems.

Endpoints

All endpoints accept POST requests with multipart/form-data as the body. This data should be just one text/plain file. You can hit this endpoint using requests like this:

import requests

local_filepath = "./path/to/local/file.txt"
url = "https://boone-rosalind.azurewebsites.net/problems/dna"

with open(local_filepath, 'rb') as file:
    files = {"file": file}
    response = requests.post(url, files=files)
    print(response.json())

If you have the dataset as a string, you can use StringIO to create a file-like object:

import requests
import io

contents = "AGCTTTTCATTCTGACTGCAACGGGCAATATGTCTCTGTGTGGATTAAAAAAAGAGTGTCTGATAGCAGC"
url = "https://boone-rosalind.azurewebsites.net/problems/dna"
file = io.StringIO(contents)
files = {"file": file}
response = requests.post(url, files=files)
print(response.json())

You can see the format of the file that each endpoint accepts in each endpoint description. The formats match the datasets that Rosalind provides. With a 200 status code, each endpoint will respond with the following json:

// a string
{
    "answer": "ACGT"
}

// or a number

{
    "answer": 0.5
}

You can write the contents of response.json()["answer"] to a file, and that should be the answer that Rosalind expects.

Testing

Run:

pytest

in the root of this project.

I bootstrap the tests by getting datasets from /app/tests/data and from the docstrings for each path. The output is checked against the corresponding output in /app/tests/output and the docstring Sample Output.

I don't do any "smart" testing (e.g. test if floats are close to each other). Instead of writing a specific test for each endpoint I thought it was easier to use the datasets provided by Rosalind. To deal with the float issue, I round all floats in the outputs to 4 decimal places and then truncate to 3 decimal places. That way I can compare the rounded floats found in the docstrings to actual results. I don't want to do any rounding when I return from my endpoints.