Could write a long answer, but I think this pretty much sums it up...

http://en.wikipedia.org/wiki/K-means_clustering

Short explanation and tutorial in R is available here:

http://www.r-bloggers.com/k-means-clustering-from-r-in-action/

and more general:

http://www.onmyphd.com/?p=k-means.clustering

There are also great web lectures by Andrew Moore available.

Dear Malgorzata , Thank U so much !! :-)

Kishna,

what kind of software you use for computations? If use R for example there are tutorials for machine learning (included k-means clustering) with R examples. For Matlab I am sure there are similar ones.

Basically k-means is about distances between variables. It could be Euclidean distance but there are other available. In R for example user chose type of distance and number of clusters.

another great lecture from slideshare, about machine learning:

http://www.slideshare.net/pierluca.lanzi/machine-learning-and-data-mining-00-course-introduction?v=default&b=&from_search=3

Dear Malgorzata,

I'm using Maestro Schrodinger /CANVAS program ( a molecular modeling software) for clustering.

In a nutshell, it is a family of methods to take a set of data points and partition (cluster) them into k different categories. Each datum D can be function of many variables, for example a function of 3 variables would be D = D(x,y,z). In a sense, the algorithm looks at these spatially (like points in space) finds subsets of them that minimize the within-cluster variance of some measure, often the distance between points. There are many other methods of clustering too...this is just one of them.