Python multidimensional dict()

Question

How would I make a dictionary that has keys that represent not just a single string, but multiple strings (array)? So each key is tied to an array that I can keep appending new strings to if the key is present in another file?

I am a relatively new Python programmer so I can visualize how my script would work... but alas I can't think of the proper syntax.

I have the logic down (I think), I just need help with which modules or functions to use in Python. I am about 3/4 the way through Learn Python the Hard Way by Zed Shaw, if you need a reference to my current knowledge in Python.

Lets get to it. Example:

File 1:

A
B
C
D

File 2:

Ted A
Mike A
Wilma B
Frank C
Dog D
Fred D

File 3:

Jon Ted
Sid Ted
Mic Mike
Will Dog
Tod Dog

Hopeful Result (write to File 4):

>A Ted Mike Jon Sid Mic
>B Wilma
>C Frank
>D Dog Fred Will Tod

So using file 1 for the keys we can fill the elements with file 2 and 3 if the key is present. Notice in file 2 A and D have multiple answers. Next look into file 3 and we expand even further; we also see that Jon and Sid(file3) = Ted(file2) = A. A also = Mike (file2) = Mic (file 3)

This is the simplest example of my data (100,000's of sequences), and the real kicker is I have five hierarchically expanding files, not just 2 (example above).

~~~ Answered ~~~

Thank you F.J and others

Try looking at `collections.defaultdict` and maybe read about [autovivification](http://en.wikipedia.org/wiki/Autovivification#Python). — JBernardo, Nov 30 '12 at 21:58
you just set a dictionary's element to an array e.g. a['foo']=[1,2,3] — Dmitry B., Nov 30 '12 at 21:59
I'm not 100% sure, but also consider looking at a digraph http://en.wikipedia.org/wiki/Directed_graph - as it sounds like you're heading in that direction... — Jon Clements, Nov 30 '12 at 22:05
@Jon Yes sir, that will be where the data goes. FYI, the data is ~100,000 sequences from 30 different "MIDs" (samples), which represent 4 different lakes. All of which can be boiled down to 15 OTUs (operational taxonomic unit). The OTUs are represented by the A,B,C above while the sequences(ID's) are represented by the names. The lakes names can be back converted from the sequence ID's. I just need to make a nice script to keep track of all of them :) — jon_shep, Nov 30 '12 at 23:25

Andrew Clark · Accepted Answer · 2012-11-30T23:21:15.227

1

The following should work:

keys = []
data = {}
names = {}

for line in open('file1', 'r'):
    keys.append(line.strip())
    data[line.strip()] = []

for line in open('file2', 'r'):
    name, key = line.strip().split()
    if key in data:
        data[key].append(name)
        names[name] = key

for line in open('file3', 'r'):
    nickname, name = line.strip().split()
    if name in names:
        data[names[name]].append(nickname)

with open('file4', 'w') as file4:
    for key in keys:
        file4.write('>{0} {1}\n'.format(key, ' '.join(data[key])))

edited Nov 30 '12 at 23:21

answered Nov 30 '12 at 22:01

Andrew Clark

202,379
35
273
306

Could you help me with the final print code in my newest edit? I just wanna make sure all the data is correctly in the hash. If true, I'll thank you for your answer here, could you also explain the line data = {}?? is that the same as saying data = dict() – jon_shep Nov 30 '12 at 23:16
@jon_shep Yes, `data = {}` is the same thing as `data = dict()`. For more info, check out this [question](http://stackoverflow.com/questions/6610606/dict-literal-vs-dict-constructor-any-preferred). – Nathan Dec 01 '12 at 00:03
So how can I learn more of this "short hand" syntax? What does the key = [] mean then? – jon_shep Dec 01 '12 at 00:10
`[]` creates an empty list, you find this in the Python [data model](http://docs.python.org/2/reference/datamodel.html) and [expressions](http://docs.python.org/2/reference/expressions.html) documentation. – Andrew Clark Dec 01 '12 at 02:03

Python multidimensional dict()

1 Answers1