Parses a cadnano file and prints the [helix, index] for each base in all strands.

parseJson

#!/usr/bin/env python
# encoding: utf-8
"""
parseJson.py

Created by Shawn M. Douglas on 2011-12-06.
"""

import sys
import os
import json

filename = "nanorobot.v2.json"

def main():
    file = open(filename, 'r')
    lines = file.readlines()
    str = ""
    for line in lines:
        str += line
    obj = json.loads(str)
    strands = obj["vstrands"]
    name = obj["name"]

    # create dictionaries (keyed by virtual helix #) of
    # row/col, scaf array, stap array
    vhToScaf = {}
    vhToStap = {}
    vhNums = []

    for strand in strands:
        num = strand["num"]
        vhNums.append(num)
        # row = strand["row"]
        # col = strand["col"]
        scaf = strand["scaf"]
        stap = strand["stap"]
        vhToScaf[num] = scaf
        vhToStap[num] = stap

    # extract scaffold 5' endpoints
    scafEnds = []
    for vh in vhNums:
        scaf = vhToScaf[vh]
        for i in range(len(scaf)):
            base = scaf[i]
            if (base[1] == -1) & (base[3] != -1):
                scafEnds.append([vh, i])

    # extract staple 5' endpoints
    stapEnds = []
    for vh in vhNums:
        stap = vhToStap[vh]
        for i in range(len(stap)):
            base = stap[i]
            if (base[1] == -1) & (base[3] != -1):
                stapEnds.append([vh, i])

    # extract scaffold paths, starting at 5' endpoints
    scafPaths = []
    for scafEnd in scafEnds:
        path = []
        [curr_vh, curr_base] = scafEnd
        [next_vh, next_base] = vhToScaf[curr_vh][curr_base][2:4]
        while next_base != -1:
            path.append([curr_vh,curr_base])
            [curr_vh, curr_base] = [next_vh, next_base]
            [next_vh, next_base] = vhToScaf[curr_vh][curr_base][2:4]
        path.append([curr_vh,curr_base])  # append last base
        scafPaths.append(path)

    # extract staple paths, starting at 5' endpoints
    stapPaths = []
    for stapEnd in stapEnds:
        path = []
        [curr_vh, curr_base] = stapEnd
        [next_vh, next_base] = vhToStap[curr_vh][curr_base][2:4]
        while next_base != -1:
            path.append([curr_vh,curr_base])
            [curr_vh, curr_base] = [next_vh, next_base]
            [next_vh, next_base] = vhToStap[curr_vh][curr_base][2:4]
        path.append([curr_vh,curr_base])  # append last base
        stapPaths.append(path)

    print "Scaffold Strands"
    for i in range(len(scafPaths)):
        print i, scafPaths[i]

    print "Staple Strands"
    for i in range(len(stapPaths)):
        print i, stapPaths[i]

if __name__ == '__main__':
    main()