https://github.com/audy/clump

Another class project
https://github.com/audy/clump

Last synced: about 21 hours ago
JSON representation

Another class project

• Host: GitHub
• URL: https://github.com/audy/clump
• Owner: audy
• License: mit
• Created: 2010-10-21T17:56:51.000Z (almost 14 years ago)
• Default Branch: master
• Last Pushed: 2015-04-28T03:17:43.000Z (over 9 years ago)
• Last Synced: 2023-03-11T01:20:52.644Z (over 1 year ago)
• Language: Python
• Homepage: http://audy.github.com/clump
• Size: 172 KB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: readme.md
  • License: license.md

Awesome Lists containing this project

README

        
# Clump

Clump is a term derived from the latin _clumpus_ which means to shake one's

belly fat in an amusing way before finding treasure.

[Austin G. Davis-Richardson]([email protected])  

## Goal

To investigate the usefulness of various clustering algorithms and distance

metrics on a dataset made from Illumina sequencing reads and classified with the SEED database into ontology subsystem trees.

## Requirements

To run these scripts you need:

   - [MatPlotLib][9]

   - [Numerical Python][8]

   - [Scientific Python][10]

   - [PyCluster][6]

Getting all of these packages installed on your system can be tricky. I recommend first installing [Python 2.6.6][5] from source into your `/usr/local/bin` directory and setting your environment variable `PYTHONPATH` to `/usr/local/lib/python2.6/`. Then, download and install all of the above packages from source (do not use the installers). My success using this method varied even on systems with the same OS and compiler versions.

If I find my methods to be useful, I may release this project in the form of a VirtualBox image to make things easier.

## Clustering

I am testing the algorithms available in PyCluster, which is a python wrapper for the C clustering library.

### Distance Computation

- Pearson Correlation

- Absolute Pearson Correlation

- Uncentered Pearson Correlation

- Absolute uncentered Pearson Correlation

- Spearman's Rank Correlation

- Kendall's T

- Euclidean distance

- City-block distance

### Clustering Algorithms

- Hierarchical clustering

- _k_-means clustering

- Self-organizing maps*

- Principal Component Analysis*

* _incomplete_

## Graphing

- Log ratio

- MA graph

- y = cases, x = controls (plain, linear)

## Dataset

These scripts take a dataset that looks like this:

If there is a `#` present in the system annotation, the gene counts are overlooked.

    # 1	i+1	..	n	Ontology

    100	20	..	90	System

    50	10	..	30	System; Subsystem 1

    50	10	..	60	System; Subsystem 2

    25	1	..	50	System; Subsystem 2; Subsubsystem 1

    25	9	..	10	System; Subsystem 2; Subsubsystem 2

(The numbers are read counts similar to expression data on a microarray,

notice they add up)

  

## Relevant Papers

Just for starters...

  - Hartigan, J. A.; Wong, M. A. (1979). "Algorithm AS 136: A K-Means

    Clustering Algorithm". Journal of the Royal Statistical Society

  - Ward, Joe H. (1963). "Hierarchical Grouping to Optimize an Objective

    Function". Journal of the American Statistical Association

  - Jain, MN Murty, PJ Flynn (1999). "Data clustering: a review". ACM

    computing surveys

  - Moon, T.K. (1996). "The Expectation Clustering Algorithm". Signal

    Processing Magazine, IEEE    

  

[1]: http://en.wikipedia.org/wiki/K-means_clustering

[2]: http://en.wikipedia.org/wiki/Hierarchical_clustering

[3]: http://en.wikipedia.org/wiki/Expectation-maximization_algorithm

[4]: http://www.theseed.org

[5]: http://www.python.org

[6]: http://pypi.python.org/pypi/Pycluster

[7]: http://bonsai.hgc.jp/~mdehoon/software/cluster/cluster.pdf

[8]: http://numpy.scipy.org

[9]: http://matplotlib.sourceforge.net/

[10]: http://www.scipy.org