Announcements |
3 Aug 06: Stepgram 2.12, with some
small bug fixes, is now available for
download
|
Input |
Input to Stepgram is a tab-delimited txt file
containing information about aCGH measurements of n probes over
m samples.
The file should contain a matrix with (n+1) rows and (m+4) columns,
according to the following format:
ProbeID |
Symbol |
Chromosome |
Position |
Sample1 |
Sample2 |
IMAGE:322807 |
W15460 |
1 |
786116 |
-0.578 |
-0.572 |
IMAGE:190915 |
H39221 |
1 |
940761 |
0.882 |
0.338 |
IMAGE:742132 |
AA406019 |
1 |
989011 |
0.032 |
NaN |
Sample dataset (4 breast cancer cell lines,
Pollack et al,
PNAS 02) (350K)
Notes
- First two columns may contain any textual information about the
probes. They are not used by the application and are copied to the
output.
- Probes must be sorted by genomic order
- Chromosome and position values must be integers. Use
23/24 to designate chromosomes X/Y (or other appropriate integers
for non-human data)
- Sample names may contain any textual string. They are copied to
the output.
- Data values should be in log-ratio units. Use NaN
or NA to denote missing values
|
Parameters |
Stepgram uses 2 parameters for aberration calling:
-
Threshold - This is the main parameter of the
algorithm, determining the sensitivity of the aberration calling
recursive procedure.
The threshold value is given in units of std, which is calculated as the
derivative log-ratio spread (DLRS), or noise level, of the data across
the entire genome. A low threshold value will result in a larger number
of detected aberrations than a high value.
Stepgram calls aberrations by searching the data vector for the most
significant aberration, then searching recursively to the left and right
for the next most significant aberration etc. The recursion is
terminated when the score of the most significant interval does not pass
the given threshold value.
Typical threshold values are in the range 6-10.
-
MinDiff - This is a secondary parameter which may
be used to tweak the aberrations calls. A common phenomenon that is
observed in aCGH data is a drifting baseline. This may result in false
detection of low-level aberrations or splitting of aberrations into
several subintervals with close levels. The minDiff parameter specifies
the minimum permissible difference between the level of an aberration
and the baseline, or between neighboring aberrations.
If your data contains a drifting baseline, set the minDiff parameter to
a value at which level aberrations should not be called (e.g. 0.1-0.2
for log2 based data). Otherwise, set minDiff to0.
Select Output penetrance to include penetrance
information in the output file. |
Output |
Stepgram outputs two tab-delimited text files containing
two different formats of the aberrations calls:
Sample |
Chr |
Beg |
End |
Probes |
Level |
1 |
1 |
23156761 |
146685011 |
260 |
-0.15072 |
1 |
1 |
156822356 |
171858339 |
50 |
0.404553 |
2 |
17 |
35127102 |
35428880 |
3 |
2.502 |
-
filename.step.txt - A file in a similar
format to the input data. The file contains the aberration level for
each probe/sample (or 0 if no aberration was detected) in place of
the measured value. This file can be easily plotted against the
original data to obtain a step function representation of the
results (see Examples section).
Optionally, the file may also include penetrance values denoting the
number of samples that contain and amplification/deletion at each
probe location (select the output penetrance option to obtain
this information).
See sample interval
output file (5K) and step output file
(440K) obtained from the sample input file
(350K) with threshold=8, minDiff=0.1 .
|
Interface |
Stepgram application interface:
- Browse: Select input data
- Parameters: Set threshold, minDiff, output penetrance
parameters
- Status: information about the status of the
application
- Calculate: Perform aberration calling analysis
(creates output files)
- Help: Open this page
- Exit
|
 |
|
References |
If you choose to use this
application in your research, please cite:
|
Contact Info |
For additional Information
please contact Doron Lipson, CS Department,
Technion, Israel.
|