ArrayMiner
Clustering
         

"The only useful clustering is one that identifies the biology behind data. Comparing ArrayMiner® to our industry-standard clustering software, we found that ArrayMiner can distinguish clusters that we know should be different according to our biological experiments. Our standard tool could not distinguish these clusters. The ArrayMiner clusters were also much more refined. In all cases, where there was disagreement, I agreed with ArrayMiner's classification."

 

Dr. Jurg Bahler,
Head, Pombe Post-Genomics, The Wellcome Trust Sanger Institute.
 
 

"GlaxoSmithKline agreed to a world-wide deployment of ArrayMiner® by Optimal Design."


ArrayMiner
 
    The ArrayMiner clustering module is an analysis tool that detects groups of coexpressed genes. Unlike most other clustering tools, ArrayMiner is a rigorous optimization tool, which means that it finds the best possible clusters, instead of using a simple algorithm and supplying a suboptimal classification. This implies that no important similarity between genes goes unnoticed, and no bogus clusters are produced.

    ArrayMiner is available as a stand-alone application and can communicate seamlessly with GeneSpring® by Silicon Genetics.

ArrayMiner algorithm
 
    Unlike the clustering methods available in most tools based on distances (like K-Means, SOM, etc.), ArrayMiner offers a rigorous statistical approach based on the Gaussian mixture model. A unique additional feature is its capacity to detect outliers (noise that does not belong to any cluster).
    ArrayMiner puts a premium on solution quality, insuring that your research time is not wasted on bogus clusters. Our proprietary grouping genetic algorithm technology insures that the huge computations are still done in short time.

ArrayMiner unique interface
 
A rich palette of views helps you gain an intimate knowledge of your data and their structure.
  Clustering interface
The main window is fully designed to give you an intuitive feeling of your data. Multiple highly customizable views are accessible, such as
  • Heat Map
  • Profile view
  • Extremely fast 2D and 3D vizualizer
Click the image to enlarge

  Comparing clusters
Unlike other tools, ArrayMiner is designed to help you understand the relationship between all your clusters. For this purpose a special interface has been designed to le you in a glance compare two or more clusters efficiently.
Click the image to enlarge

  Comparing classifications
The classification compare window lets you analyse the differences between two or more classifications. It lets you for example analyse how the classification behaves when you increase the number of clusters. Or easily compare your clustering result with a known biological classification.
Click the image to enlarge

  Building experiment and clustering trees
In addition to ArrayMiner's unique data mining algorithm, an experiment tree window has been added. The Experiment tree module gives you an intuitive way to see how your experiments or your clusters match each other.
Click the image to enlarge

  Links
  • Discover ArrayMiner classmarker here
  • Lean more about ArrayMiner algorithm here
  • Discover the publishing tools of ArrayMiner here
  • View Quicktimer movies showing ArrayMiner usage here
  • Ask for a fully functional demo to try ClassMarker on your own data here