Hello all,
I am cross posting this from stackoverflow as I was did not receive good response or it was not well received,
I have just started to learn programming using python and hence asking a pythonic way (not pandas as it is very advanced) to approach this problem. any tips or comments is appreciated as it will help me learn.
http://stackoverflow.com/questions/43482716/creating-a-matrix-using-python-for-biologist
Thank you.
•
written
by
novicebioinforesearcher •
You can read/write csv files using the csv module.
import csv
from collections import defaultdict
Make a list of files you want to read
myFiles = ['file1.csv', 'file2.csv']
Go through the files and get the genes, store them in a dictionary
myDict = defaultdict(list)
for file in myFiles:
openF = open(file, 'r')
csvIn = csv.reader(openF, delimiter = ',')
for line in csvIn:
gene = line[0]
expression = line[1]
myDict[gene].append(expression)
So you will have a dictionary where the keys are the gene ids. Using the key you will get a list of expression values. You can then go through the dictionary key by key and write the output
You may not understand all this code right now but its fairly simple, look up any parts you dont understand. You may have problems if one file has a gene the other does not but you can adapt the code to fix that in multiple ways
Thanks fuzzy, i wrote something of this sort but i ran into the exact same problem you mentioned, i have some some genes which are not there in other file, how would I modify to account for genes/junctions that are not there in other files, like put a if argument if key has gene but not value then add 0?
Log in to answer this question.
Is there a reason this has to be in python? In R, it's
outer_join()from the dplyr package.BTW, in reality the genes in the files will all be present and in the same order, so you'll never have missing values.
Dear Devon,
thank you for your reply, that was my assumption to but i am dealing with gene names that have been collapsed from tophat junction bed file, I have to make a matrix of these gene names, so there could be a chance that at some places there would be no reads mapping to junction. eg
i need to do this to make a dataset that i can use for a tool where i am trying to see differential junction usage
What exactly couldn't you manage with the
pandassolution in that thread? It looks pretty well broken down to me and that's exactly how I'd go about doing it. If you're struggling to install pandas, try:python -m pip install pandas.If you don't have administrative access, consider using the Anaconda/Miniconda distributions.
While we sympathise with your novice status, there is no real substitute for just spending the time trying to figure it out - especially when you're asking for a kind of general solution to a specific task (only you know exactly what your data is going to look like each time).
hello @jrj.healey,
I am trying to learn as i go pandas seems to be way advanced, if i can do it in a non pythonic way it would make more sense to me and more over the final output i get does not match the read counts from individual files, may be it has to do with difference in each file, some files have 222000 lines where as some of them have 300000 lines.
I'm not sure what you mean by a more 'pythonic' way? The ethos of python is this:
I'd say in this case
pandasis the aforementioned 'obvious way' (with the possible exception of Fuzzy's solution withcsv). It might seem like an advanced step, but what you're basically doing is giving yourself access to many clever functions which have been prewritten to allow you to manipulate dataframes. I'd wager you'll have net saving of time investment if you spent it learning a bit ofpandasinstead of struggling throughfor,if&elsestatements.You'll need to edit your question and provide reproducible input files and commands before we can address what problems you might have had with the various solutions - there isn't really enough to go on at the moment.