JUMPSTART is a software package to perform statistical analysis on MPSS measurements
Introduction
IBM JUMPSSTART is a tool that analyzes Massively Parallel Signature Sequencing (MPSS) data and provides several helpful views of MPSS data and models.
Execution
JUMPSSTART is executed from the command line. Upon successful execution it will display a GUI control panel. There are two versions available here as gzip compressed tar files: an OS X executable and a Windows 32-bit executable. The Windows zip file also contains opengl32.dll. This is the standard OpenGL library and JUMPSSTART requires it to run. If you do not already have OpenGL installed on your computer please copy the opengl32.dll to you WINDOWS\System32 folder prior to running JUMPSSTART. We also have a small demo dataset here for download.

Loading and selecting data
You can load an MPSS data file by pressing the load button. Upon doing so a GUI data selection panel will appear.

This panel shows the available datasets. You can select any number of the data columns by simply checking the box to the left of the column name.
Merging data columns
Often MPSS datasets contain multiple runs of the same stepper batch
stored as columns. You can merge these steppers runs by selecting
them in the selection panel and pressing the merge selected
after providing a name for the resulting merged data column by
typing it into the into
field right of the button. If you do
not provide a name the result of the merge will fall into the void
Modeling replicates
If you wish to model replicate data columns you do so first
selecting the two replicate columns in two pull-down menus within
the modeling commands
area, providing a name for the
resulting model in the as
field, and then pressing
the fit
button. If you do not provide a result name then
JUMPSSTART will display the default model and data views and then
allow the model to fall into the void.
Saving results
JUMPSSTART allows you to save two types of results, column-wise and
pairwise, by pressing the corresponding button in the file
commands
area. In both cases JUMPSSTART analyzes the data
currently selected in the selection panel using the model currently
selected in the current model
pull-down menu. JUMPSSTART will
use the terminal to query a file name and then writes the results
as a tab delimited text file.
column-wise results
Saving column-wise results writes a matrix containing a shared signature column followed by four result columns for each selected data column.
SIGNATURE name1 atoms zeros delta name2 ...
signature1 tpm11 atoms11 zeros11 delta11 tpm21 ...
signature2 tpm12 atoms12 zeros12 delta12 tpm22 ...
... ... ... ... ... ... ...
signatureN tpm1N atoms1N zeros1N delta1N tpm2N ...
pairwise results
Saving column-wise results writes a tab delimited matrix containing a shared signature column followed by pvalue for all pairs of the selected data columns.
SIGNATURE name1_vs_name2 name1_vs_name3 ...
signature1 pvalue1 pvalue1 ...
signature2 pvalue2 pvalue2 ...
... ... ... ...
signatureN pvalue3 pvalue2 ...
MPSS data
MPSS data is a tab delimited text file in the following format
SIGNATURE name1 name2 ... nameN
signature1 count11 count12 ... count1N
signature2 count21 count22 ... count2N
... ... ... ... ...
signatureN countN1 countN2 ... countNN
where name* are any whitespace free strings, signature* are MPSS signatures represented by strings of length 17 containing only the letters from the set [ATCG] and beginning with GATC, and count** are integer beads counts.
MPSS models
G. Stolovitzky, A. Kundaje, G.A. Held, K. Duggar, C. Haudenschild, D. Zhou, T. Vasicek, K. Smith, A. Aderem and J. Roach, Statistical analysis of MPSS measurements: application to the study of LPS-activated macrophage gene expression, Proc. Natl. Acad. Sci. USA, 102 (5), 1402-1407 (2005). (Pubmed) (PNAS site)
The void
The void
in JUMPSSTART is simply an empty space that consumes
anything that falls into it.
