Currently, we provide descrtiptions of analytical features of SegMine (i.e., SegMine workflow components) and video tutorials. Together with use case documentation and installation instruction we hope this is enough to get started with SegMine.



Microarray parser

Microarray parser widget performs three functions:
  1. parses the input data, recognizes values in rows and columns, treats missing values as specified and performs syntax check,
  2. creates a native data structure (orange.ExampleTable) from the data, and
  3. computes logFC values.
INPUTS:
  • microarrayAsString [type: string]
OUTPUTS:
  • microarrayTable [type: orange.ExampleTable]
  • logFCvalues [type: list of tuples (string, float)]

Gene ranker

Gene ranker widget ranks input genes using the ReliefFalgorithm.

It also computes t statistic for independent two-sample t-test assuming unequal sample sizes and unequal variances.

The widget returns a list of tuples (geneName, rank), sorted in descending order according to rank.
INPUTS:
  • microarrayTable [type: orange.ExampleTable]
OUTPUTS:
  • geneRanks [type: list of tuples (string, float), sorted in descending order]
  • tScores [type: list of tuples (string, float), sorted in descending order]

Rank plotter

Rank plotter widget draws a dot graph of the input data. It is typically used to inspect the rank curve. x-axis also supports infinite zoom.
INPUTS:
  • geneRanks [type: list of tuples (string, float)]

Cutoff

Cutoff widget removes those genes from the input list whose logFC values do not fit in the specified range. It provides upper and lower bound which can also be absolute. Most typically, genes with absolute logFC value around 0 are filtered out.
INPUTS:

  • geneRanks [type: list of tuples (string, float)]
  • logFCvalues [type: list of tuples (string, float)]
OUTPUTS:

  • reduced geneRanks [type: list of tuples (string, float)]
  • reduced logFCvalues [type: list of tuples (string, float)]

Resolve HMR gene names

Resolve HMR gene names widget tries to resolve symbolic names and synonyms for input genes (human, mouse, rat). It should be applied to the data just before they are put to the SEGS HMR widget.
If an input gene is not recognized, it is simply ignored.
INPUTS:
  • geneRanks [type: list of tuples (string, float)]
OUTPUTS:
  • geneRanks [type: list of tuples (string, float)]

Hierarchical clustering

Hierarchical clustering widget implements interactive hierarchical clustering. There are many options for controlling the cluster visualization. It provides the following linkage criteria:
  • single linkage
  • average linkage
  • Ward linkage
  • complete linkage.
Basically, Hierarchical clustering widget is provided by Orange but it is enhanced by Orange4WS so it also outputs the actual clustering which can be sent to the Rule browser to display cluster information.
INPUTS:
  • DistanceMatrix [type: orange.SymMatrix]
OUTPUTS:
  • SelectedExamples [type: orange.ExampleTable]
  • UnselectedExamples [type: orange.ExampleTable]
  • Centroids [type: orange.ExampleTable]
  • Structured Data Files [type: DataFiles]
  • HierarchicalClustering [type: orange.HierarchicalClustering]

Example distances

Example distances widget computes distances between examples (rows) in the input table. In SegMine scenarios, rows in the table are rules as computed by the SEGS algorithm, and columns can be individual genes or individual terms.

Different distance metrics can be used:
  • Euclidean
  • Pearson correlation
  • Spearman rank correlation
  • Manhattan
  • Hamming
  • Relief
INPUTS:
  • Examples [type: orange.ExampleTable]
OUTPUTS:
  • DistanceMatrix [type: orange.SymMatrix]

SEGS for HMR

SEGS HMR widget provides an interface to the web service implementing the SEGS algorithm for Homo sapiens, Mus musculus, and Rattus rattus. More details about the SEGS algorithm can be found in the paper and in the phd thesis of Igor Trajkovski.
INPUTS:
  • geneRanks [type: list of tuples (string, float), sorted in descending order]
OUTPUTS:
  • combinedRules [type: list of web service response objects]
  • fisherRules [type: list of web service response objects]
  • GSEArules [type: list of web service response objects]
  • PAGErules [type: list of web service response objects]
  • jobID [type: string]
  • resultsAsXML [type: string]

Fetch SEGS results

Fetch SEGS results widget fetches existing results of the SEGS algorithm from the server. There are two versions: one for SEGS HMR and one for plants.
INPUTS:
  • jobID [type: string]
OUTPUTS:
  • combinedRules [type: list of web service response objects]
  • fisherRules [type: list of web service response objects]
  • GSEArules [type: list of web service response objects]
  • PAGErules [type: list of web service response objects]
  • resultsAsXML [type: string]

Load SEGS results

Load SEGS results loads existing results of the SEGS algorithm from an XML file. There are two versions: one for SEGS HMR and one for plants.
OUTPUTS:
  • combinedRules [type: list of web service response objects]
  • fisherRules [type: list of web service response objects]
  • GSEArules [type: list of web service response objects]
  • PAGErules [type: list of web service response objects]

Rule browser

Rule browser components displays rules obtained by the SEGS algorithm in an HTML table. Rule terms as well as covered genes can sent to the widget connected to the output by clicking on the corresponding link.
Rule browser can also display rule clusters information if the Hiearchical clustering component is connected to its input. There are two versions: one for SEGS HMR and one for plants.
INPUTS:
  • rules [type: list of web service response objects]
  • HierarchicalClustering [type: orange.HierarchicalClustering]
OUTPUTS:
  • termList [type: list of strings]

ExampleTable from rules

ExampleTable from rules constructs an orange.ExampleTable structure from a list of rules, computed by the SEGS algorithm. It constructs two tables: the first one has genes as columns and the second has terms as columns.
INPUTS:
  • rules [type: list of web service response objects]
OUTPUTS:
  • RuleGeneTable[type: orange.ExampleTable]
  • RuleTermTable[type: orange.ExampleTable]

Biomine neighbourhood search

Biomine neighbourhood search widget provides an interface to the Biomine search engine. It performs neighbourhood search which searches for neighbouring entities of given query terms and returns a graph in .bmg format. There are two versions: one for HMR and one for plants.
INPUTS:
  • startNodes [type: list of strings]
OUTPUTS:
  • graphAsString [type: string]
  • bestPathsGraphAsString [type: string]

Biomine connections search

Biomine connections search widget provides an interface to the Biomine search engine. It performs connections search which searches for connections between two sets of given query terms and returns a graph in .bmg format. There are two versions: one for HMR and one for plants.
INPUTS:
  • startNodes [type: list of strings]
  • endNodes [type: list of strings]
OUTPUTS:
  • graphAsString [type: string]
  • bestPathsGraphAsString [type: string]

Add logFC to Biomine graph

Add logFC to Biomine graph widget adds logFC values to vertices representing input genes in a given Biomine graph.
INPUTS:
  • logFCs [type: list of tuples (string, float)]
  • graphAsString [type: string]
OUTPUTS:
  • graphAsString [type: string]

Biomine medoids search

Biomine medoids search widget provides an interface to the Biomine search engine. It performs medoids search which searches for medoid entities of given query terms, and returns a list of medoids as well as list of all query terms with their medoid scores.
INPUTS:
  • startNodes [type: list of strings]
OUTPUTS:
  • medoids [type: list of strings]
  • rankedTerms [type: list of tuples (string, float)]

Biomine graph painter

Biomine graph painter widget applies the specified colour to the given terms in a given Biomine graph. Using this widget interesting terms can be easily recognized in a large graph.
INPUTS:
  • termsList [type: list of strings]
  • graphAsString [type: string]
OUTPUTS:
  • graphAsString [type: string]

Biomine graph visualizer

Biomine graph visualizer widget provides a powerful interactive Biomine graph visualizer. It is distributed along with SegMine, and runs localy as a Java Applet.
NOTE: this component requires Java Runtime Environment (JRE) to be installed.
INPUTS:
  • graphAsString [type: string]
OUTPUTS:
  • graphAsString [type: string]
SegMine news
  • SegMine for metabolomics is under development.
  • SegMine for ClowdFlows is under development
  • Version 0.7.2 is out!
  • Web services used in SegMine are now hosted on a powerful new server with plenty of memory and processors.
Authors and institutions