How does a bioinformatician employ Python in data analysis?

Question

[fi1d18]

Hi,

When I am asked to mention for which I have recently used a programming language like Python in data analysis, I just can say that I used Python once to extract some genome coordinates from gene bank. But definitely this is not a wise answer and disappointing for a bioinformatician. How then a medium data analyst could use Python? I mean for which purpose as a good answer.

Thank you very much

[Devon Ryan]

If someone is asking you in what context you've used something it's best not to answer that you've used it in some context in which you haven't (i.e., don't lie). If you start mentioning things you've never done then if they ask any follow up questions you'll be immediately screwed.

[GenoMax]

Is the focus of the question to figure out if you know python (and how to code) because otherwise that question does not make a lot of sense. If you are an end-user/consumer of code (python or else) that others make so generously available, then I don't see any problem in making that clear when you answer the question.

[fi1d18]

Thank you, exactly as what you stated I am an end-user/consumer of code and as WouterDeCoster mentioned I use Python and R for differential expression, plotting and data processing. you are always helpful

Answer 1

[WouterDeCoster]

If this is hypothetical (and not explicitly about how you have used Python) then you can say that Python can be used for pretty much everything, although it has it's strong points and weaker points.

You could use python to create a pipeline (snakemake), make plots (matplotlib, seaborn), interface with databases (sqlite), perform processing of fasta/fastq (biopython) and create websites (django).

But even more important than knowing what you can is knowing the limitations. While you have a lot of statistical modules in Python - usually R is the more appropriate language here, definitely for differential expression analysis. But there is nothing that you can do in R that you cannot do in Python. For making production ready code for which speed is an issue then Python is often not your best guess, rather something compiled such as C or Java, or more recent languages such as Go or Rust. It's often 1) having a language you are comfortable in and 2) using the right tool for the job at hand.

I really dislike R, but I do differential expression analysis in R.

Answer 2

[Alex Reynolds]

You might keep an interactive notebook that is Python-based, like Jupyter, or IPython, or Google's recent Colaboratory effort, which also appears to tie in Tensorflow-based machine learning components.