The
new release introduces many new features including a first version of CGHTools for the analysis of array Comparative Genomic
Hybridization data.
1.
ANOVA plug-in on log intensities:
Added an option for evaluating pairwise contrasts.
2: A Plugin for planning the number of arrays needed
for identifying genes differentially expressed between 2 classes
3: Gene Set Comparison:
Added Efron-Tibshirani’s GSA maxmean test and Goeman's global
test. The structure of the HTML output is simplified to give a comparison of
results for all tests. The Hotelling test was
dropped.
4: BROAD gene set
Updated access to the Broad Institute Molecular Signature Database (MSigDB) including positional, curated,
motif, and computed gene sets. Added 'rat' and mouse species for the curated gene sets group.
5: Class prediction
Added a cross-validated ROC curve to the HTML output for the Bayesian compound
covariate predictor.
6:ScatterPlot:
Added a button to export the genelist. Also, modified the name of the scatterplot
for phenotype averages to reflect the class variable.
7: Clustering: Heatmap
Enhanced the heatmap to provide a color gradient
option that the user can select from a color pallet. Also,
modified the heatmap zoom-in feature to allow the
user to specify the gene identifier to be displayed.
8:Normalization for Single Channel:
Modified the code to allow the user to normalize the data by groups of arrays.
9: GO download:
Modified the code to obtain the Gene Ontology files from Bioconductor.
10: Affymetrix annotations:
Modified the code to download the Affy annotations
from Bioconductor.
11: DrugBank Link:
Added to the "info" link in the HTML output, a link to query using
gene symbol.
12: Updated the import wizard:
The automatic importers for Affy, Agilent,
Genepix and mAdb can now be
accessed using the data import wizard.
Version
3.6.0 stable release contains the following bug fixes and improvements:
1) Class Prediction: Fixed a bug in the class
prediction output where the t-statistic column had 1e-07 values instead of
negative values. Also, for the CCP and DLDA prediction methods modified
the code to handle missing values when computing the weights and threshold.
2) Random Variance Model: Increased the limit on
the number of genes to 500K as well as increased the corresponding stack size.
3) Quantitative Trait analysis: Fixed a bug where
the HTML output showed 1e-07 instead of negative values for correlation
coefficients.
4) Zoom and recolor clustering: Previously, the
user specified class column to label the experiments was not displayed but this
has been fixed in this release.
5) Survival gene set comparison: Fixed an error
that was caused when incorrectly loading the default parameter file.
6) Modified the gene index in the Fortran code to handle more than 9 digits in various
analyses.
7)
Survival Risk Prediction: Modified the tool for the special case when no genes
are selected in the combined model such that when cross-validating the model
the gene with the smallest p-value together with clinical covariates will be
used in the Cox regression and prediction.
1) The system architecture
has been modified in this version of BRB-Arraytools
to handle more than the Excel limit of 65,000 rows. The gene identifier and
gene annotation information in now stored binary files.
2) This version of BRB-ArrayTools is compatible with MS Vista and Excel 2003.
3)
Data Import
GEO
importer: This tool allows users to automatically import a GDS dataset from the
NCBI Gene Expression Omnibus (GEO) database into BRB-ArrayTools.
Agilent importer: The data import wizard now automatically recognizes
the format for dual channel Agilent data and directly
imports the background subtracted intensities and annotations.
Affymetrix .CEL files: (i) For large number of .CEL files (greater than 100), to avoid
memory problems, we have implemented a new method called ‘almostRMA’. This method uses a subset of arrays to
compute the quantile normalization and probe effects
model and then applies these to all the arrays in the data set. (ii) A new
option to compute MAS5.0 probe set summaries from .CEL files has been included.
4)
Analysis Tools:
Gene
Set Expression Comparison: We created two new families of gene sets that can be
used within the Gene Set Expression Comparison tool. One family contains the
set of genes that are targets of a transcription factor; one set for each TF,
with the option to use experimentally verified targets or computationally
determined putative targets. The second family contains a set of
computationally determined putative targets for each microRNA.
Survival
Gene set Expression Analysis: This analysis tool finds sets of genes for which
the expression levels are correlated to survival. Similar to the Gene Set
Expression comparison tool, this tool can be used to analyze Gene Ontology
categories, Pathways, micro RNA targets, transcription factor targets and user
defined gene lists.
Enhanced
plug-in ANOVA of log intensities: This enhanced plug-in replaces the Class
comparison tool between Red and Green channels. The plug-in is used for finding
genes differentially expressed between two classes for two-color arrays without
a common reference sample. It can also be used to compare samples of one class
with the reference samples in the common reference design.
Class Prediction: We have
implemented a new option for gene selection based on recursive feature
elimination. The user specifies the number of genes to include. Starting with a
full model the method excludes genes whose correlation with outcome is minimal.
This reduction continues until the target number of genes is reached. The
recursive feature elimination is applied from scratch within each
cross-validated training set. Although recursive feature elimination is based
on a support vector machine model, any type of classifier can be used for the
genes selected for the training set.
Bayesian compound covariate
predictor: We added an option of not predicting any class if the greatest
posterior probability does not exceed a user-specified threshold. The HTML
output now also displays the predicted probability.
5) We provide a new utility to
create and save for further analysis a list of genes that are correlated to a
user-specified gene based on a user-specified threshold.
6) We modified the format of
the genelists that get generated from an analysis tool
to include gene annotation information whenever available. This facilitates use
of such gene lists with data from different projects or with different
platforms.
7) This version has the
capability to simultaneously run more than one analysis tool within a project.
Version
3.5.0 stable release contains the following bug fixes and improvements:
=====================================================
1)
Gene Set Expression Comparison: Fixed a bug that occurred when the Random
Variance Model option was selected; it was not used in the analysis of GO
categories and Pathways.
2)
False Discovery Rate(FDR): The False discovery rate
reported in the HTML output has been corrected. The magnitude of difference to
the previously reported FDR values appears small (e.g
10^(-2)).
3)
Broad/MIT Pathways: Modified the code to accommodate for the changes made on
the Broad/MIT web page. Enhanced the HTML output by providing
hyper-links for some of the gene sets.
What's New
in BRB-ArrayTools Version 3.5 – Beta Release
====================================================================
1)
Data Import Wizard
We
have added a new data import wizard to assist users through the process of
importing their data into BRB-ArrayTools.
2)
GC-RMA
In
this version, the GC-RMA method for computing probe set summaries from .CEL
files has been implemented.
3)
Analysis Wizard
An
analysis wizard has been added to guide users with selecting the appropriate
analysis tool based on their research question and experimental design.
4) Survival
Risk Prediction
We
have enhanced the Survival Risk Prediction tool to allow up to 3 risk groups
and 3 clinical covariates.
5)
Class Prediction
A
new method called the ‘Bayesian Compound Covariate predictor’ has
been included for two classes. The Top Scoring Pair class prediction plug-in
has been extended to use multiple pairs of “synergistic” genes.
Additionally, for the greedy pairs option we have
enhanced the output to include the gene pair information.
6)
0.632+ bootstrap Cross-validation
The 0.632+
bootstrap method of cross validation replaces the 0.632 method for estimating
prediction error.
7)
Gene Set Expression Comparison
An
additional test called the Hotelling’s T square
test, has been implemented for finding significant
gene sets.
8) Affymetric Quality Control Plots for .CEL files
We
have added a utility to provide quality control plots and RNA degradation plots
for projects imported using Affymetrix CEL files.
9)
Clustering
We
have improved the color scale for the heatmap in BRB-ArrayTools. We have also added an option to median center
single channel data when using the Cluster 3.0/Treeview tools.
10)
Preferences
Added a preference menu option to allow users to modify certain
preference parameters for BRB-ArrayTools.
11)
Log File
A new
log file has been added which records the parameter options used at the time of
importing the data and performing an analysis.
12)
Mac Users
This
version has been successfully tested with Windows XP professional running on
Apple macbook pro machine. The windows XP
professional was installed with Apple’s bootcamp
software.
Version
3.4.0 stable release contains the following bug fixes and improvements:
1)
Random Variance Model: Modified the code to handle missing values consistently in
various tools.
2)
Time Series Plug-in: Modified the code to pass the 'time' variable as a numeric
instead of a factor.
3)
Class Prediction: Fixed the SVM error message when the optimization process did
not end.
4)
Quantitative Trait: Corrected the HTML output by removing Global test p-value
from the HTML output.
5)
Horizontally aligned file format: Fixed the run time error caused due to long
header lines.
6)
Click to display: Fixed an error on the "Filtered log
ratio/intensity" worksheet to use a numeric sort.
7)
Non-English Language Users: Implemented a fix for non-English language users
What's New
in BRB-ArrayTools Version 3.4 – Beta Release
1)
Re-designed the architecture of BRB-ArrayTools so
that there is no longer any restriction on the number of arrays that a project
can contain.
2)
Survival Risk Group Prediction
3)
Gene Set Expression Comparison Using Broad/Whitehead Signatures and Pathways
4)
Create User Defined Gene List Based on GO Terms
5)
Top Scoring Pairs Class Prediction
6) BRB-ArrayTools Archive of Human Tumor Projects
7)
Improvement of User Dialogs
Version
3.3.0 Stable Release contains the following bug fixes:
1)
Added an option for background adjustment,fixed
bug for reverse fluor in GenePix
Importer.
2)
Fixed the bug using CGAP,Biocarta
and KEGG pathways in Gene Subset.
3)
Fixed the bug in print-tip Lowess normalization,using housekeeping genes for single channel.
4)
Class Prediction: Progress bar now works when permutation test option is
selected. Also, corrected the expression data table in HTML output.
5)
PAM: Added a warning message about the impute function.
6) Plugins:The following plugins may have passed incorrect data. For 1-color data:
Histogram and Smoothed CDF and 2 color data:ANOVA on log intensities, Histograms, Pairwise Correlation Plot, MA plot and Smoothed CDF.
What’s
New in BRB-ArrayTools 3.3 beta
1)
Enhanced heat map
more color coding options
including multi-color rainbow
zoom in and out
labeling of genes
2)
Pathway annotation of gene lists
3)
Class comparison based on pathways rather than individual genes
4)
Fast Fortran implementation of SAM
Approximately 7x faster than other implementations
5)
Normalization of data separately by grid (print tip) for printed arrays
6)
Direct import of GenePix data
7)
Enhancements to Class Prediction analysis
Optimization of significance threshold for gene selection
New algorithm for selecting effective pairs of genes
Addition of shrunken centroid (PAM) classifier
8)
New re-sampling methods for estimating prediction error
K-fold repeated cross-validation and .632 bootstrap options
9)
Utility to compare gene lists
10)
Plug-in for
11)
Plug-in for regression analysis of time series data to find regulated and
differentially regulated genes
Version
3.2.1
=============
Version
3.2.1 is the stable release of version 3.2. Various bug fixes have been
made. For details, please see the Readme file.
What's New
in Version 3.2
=========================
1)
Automatic importation of Affymetrix CEL files.
Calculation of Affymetrix probe set summaries and
normalization using RMA function of Bioconductor.
2)
Importation of either log-transformed or not log-transformed data.
3)
Class comparison to determine significance of Gene Ontology categories.
4)
Class comparison to determine significance of user-defined genelists.
5)
Extended class comparison for use in red-to-green comparisons with common
reference.
6)
Significance Analysis of Microarrays (SAM).
7)
Speeded-up binary tree prediction tool using K-fold cross-validation.
8)
ANOVA plugins tools:
-
Fixed effect model for log-ratio or log-signal with up to 4 factors.
-
Random effects model for log ratio or log signal.
- ANOVA
for single channel intensities for dual-label arrays using non-common-reference
design.
9)
Optional parameters in 'Preferences.txt' file (in Prefs folder of ArrayTools
installation folder) to control size of dendrogram
plots produced by clustering tools.
10)
Sample statistical considerations sections for publications included in Help
documents.
What's new
in release 3.1
=================================
1)
New collation wizard simplifies the importation of expression data.
2)
Automatic importation of Affymetrix annotations.
3)
Data can now be directly ported to Cluster and TreeView
for further analysis.
4)
New class comparison tool for dual-label arrays that do not use common
reference design.
5)
New quantitative traits analysis tool allows selection of genes which are
significantly correlated with a quantitative (numeric) trait.
6)
New binary tree prediction tool for class prediction with more than two classes
by forming a tree of binary classifiers.
7)
The class prediction tool has two new options for the gene selection criteria
and the randomized variance method has now been implemented as an option in the
class prediction tool.
8)
Improved gene ontology analysis of gene lists produced by class comparison,
class prediction, survival and quantitative traits tools.
9)
Users may now subset the data globally for any analysis by selecting a set of
defined genelists.
10)
If the user is connected to the Internet, then BRB-ArrayTools
will automatically check the BRB server for software updates every seven days
(unless the user changes the UpdateInterval parameter
in the Preferences.txt file).
Changes
since the previous version (3.0.2e):
============================================
1)
BRB-ArrayTools now uses version 1.2 of the
RCOM. If the user does not yet have the RCOM 1.2 installed, then the BRB-ArrayTools installer will automatically try to install
it. The user may still use any version of R later than or equal to
version 1.6.1.
2)
The collation wizard now offers the option of whether or not to average over
multiply-spotted clones during collation. Previous versions automatically
averaged over multiply-spotted clones when the original input data were in
separate files for each array, but did not do the averaging when the original
input data were horizontally aligned in one single file.
3)
The order of appearance of the arrays in the "Filtered log ratio" or
"Filtered log intensity" worksheet of the project workbook now
matches the order of appearance of the arrays in the experiment descriptor
file, if the user has provided one.
4)
Permutation tests now use a complete enumeration of permutations when the
number of possible permutations is less than 1000, and only uses randomly selected
permutations when the number of permutations exceeds 1000.
5)
The default confidence interval for the multivariate permutation test is now
90%.
DOWNLOAD
NOW http://linus.nci.nih.gov/BRB-ArrayTools.html