Dynamic and Interactive Visualization
The platform offers powerful visualizations that are interactive and allow dynamic changes. This provides an easy option of visually mining the data and reducing its dimensionality.
The data views provided are the Spreadsheet, the Scatter Plot, the 3D Scatter Plot, Profile Plot, Heat Map, Histogram, Profile Match, Spot Image, Annotation View, Matrix, Summary Statistics, and Bar Chart.
All the active views are lassoed, i.e., selections on one view are propagated to all other open views.
Unsupervised Machine Learning
Cluster analysis is a powerful way to organize datasets into groups or clusters of similar profiles. There are several ways of defining the similarity measures, or the distances. While some methods are purely mathematical, others use domain specific knowledge about the data.
The platform has several clustering algorithms.
The algorithms provide multiple options of distance functions like Euclidean, Square Euclidean, Manhattan, Chebychev, Differential, Pearson Absolute and Pearson Centered.
Data is sorted on the basis of such distance measures to group both rows and columns into most similar clusters. Since different algorithms work well on different kinds of data, this large battery of algorithms and distance measures ensure that a wide variety of data can be clustered effectively.
The results are displayed as interactive views such as the Cluster Set View, the Dendrogram View and the Similarity Image View.
Supervised Machine Learning
Classification algorithms are a set of powerful tools that allow researchers to exploit experimental data for learning-based prediction of outcomes. In the platform, classification comprises a set of supervised learning algorithms, which construct a model from training dataset. This model is used to predict classes for new unclassified data.
Model building for classification is done using four powerful machine learning algorithms:
Models built with these algorithms can be used to classify samples into discrete classes. The models built by these algorithms range from visually intuitive (DTs) to very abstract. Together, these methods constitute a comprehensive toolset for learning, classification and prediction.
The platform also offers regression algorithms like
Statistics and Hypothesis Testing
Hypothesis testing can be performed on any dataset in which columns correspond to experiments and rows to the endpoints being measured. Hypothesis testing will produce a significance p-value for each row.
The platform offers variety of statistics including:
The platform is also integrated into the 'R' statistical package offering a host of statistical functions.
Data Import Wizards
Multiple forms of data ranging from text files to chemical structures to proprietary Microarray formats can be imported with ease.using these wizards.
Normalization & Data Pre-processing
The platform offers three kinds of normalization Mean/Median Shifting, Linear and Non-Linear Lowess Regression and Quantile Normalization for Replicates. For specific cases in microarray data analysis, normalization using spike-in genes or housekeeping genes is also available.
Biological Annotations
After desired information is mined from experimental data, it is very useful to curate related information from various sources/repositories. For gene expression data analysis applications, the platform provides options for annotating the genes of interest from various public sources like EntrezGene, Unigene, LocusLink, PubMed etc. It also provides gene ontologies and pathways information. |
Solutions 
