about: wrong information of CN 500k

[haile...@gmail.com]

Hi Henrik,

I am analyzing 500k SNP array, I just followed the steps provided on your webpage. (I did not use quantileNormalization, in stead of AllelicCrosstalkCalibration) But When I run PCR fragment length normalization, I got the follwoing wrong information:

Normalizing set for PCR fragment-length effects...
Identifying SNP and CN units...
types
2
238378
subsetToUpdate:
int [1:238378] 1 2 3 4 5 6 7 8 9 10 ...
Identifying SNP and CN units...done
Retrieving SNP information annotations...
UflSnpInformation:
Name: Mapping250K_Sty
Tags: na24,HB20080214
Pathname: annotationData/chipTypes/Mapping250K_Sty/Mapping250K_Sty,na24,HB20080214.ufl
File size: 466.04kB
RAM: 0.91MB
Chip type: Mapping250K_Sty
Number of enzymes: 1
Retrieving SNP information annotations...done
Identifying the subset used to fit normalization function(s)...
int [1:233558] 1 2 3 4 5 6 7 8 9 10 ...
Identifying the subset used to fit normalization function(s)...done
Shift: 0
onMissing: median
Array #1 of 338 ('EA06028_0091-02A_Sty250_SS247725@PCRS')...
V1
Min. : 115
1st Qu.: 440
Median : 598
Mean : 609
3rd Qu.: 771
Max. :1266
NA's : 149
int [1:233558] 1 2 3 4 5 6 7 8 9 10 ...
Estimating target prediction function...
Get average signal across arrays...
Retrieving average cell signals across 338 arrays...
CnChipEffectFile:
Name: .average-intensities-median-mad
Tags: 6be84633e00f19d0094dc5127334bf17
Pathname: plmData/snp_all,ACC,-XY,RMA,A+B/Mapping250K_Sty/.average-intensities-median-mad,6be84633e00f19d0094dc5127334bf17.CEL
File size: 8.70MB
RAM: 0.02MB
File format: v4 (binary; XDA)
Platform: Affymetrix
Chip type: Mapping250K_Sty,monocell
Timestamp: 2008-11-06 09:44:50
Parameters: (probeModel: chr "pm", mergeStrands: logi TRUE, combineAlleles: logi TRUE)
Retrieving average cell signals across 338 arrays...done
Get average signal across arrays...done
used (Mb) gc trigger (Mb) max used (Mb)
Ncells 1474426 78.8 4574153 244.3 7147115 381.7
Vcells 9107142 69.5 29133235 222.3 60960256 465.1
Units:
int [1:233558] 1 2 3 4 5 6 7 8 9 10 ...
Getting (unit, group, cell) map...
Retrieving unit-to-cell map...
Found (unit,group,cell) map cached on file
Retrieving unit-to-cell map...done
Getting (unit, group, cell) map...done
Group 1 (1) of 1...
Group 1 (1) of 1...done
Unit-by-group cell matrix map:
'UnitGroupCellMatrixMap' int [1:233558, 1] 1 2 3 4 5 6 7 8 9 10 ...
Thetas:
num [1:233558, 1] 0 0 0 0 0 0 0 0 0 0 ...
(Allele-specific) thetas:
num [1:233558, 1] 0 0 0 0 0 0 0 0 0 0 ...
Total thetas:
num [1:233558] 0 0 0 0 0 0 0 0 0 0 ...
Summary of total signals (on the intensity scale):
Min. 1st Qu. Median Mean 3rd Qu. Max.
0 0 0 0 0 0
Signals:
num [1:233558] -Inf -Inf -Inf -Inf -Inf ...
Summary of signals (on the log2 scale):
Min. 1st Qu. Median Mean 3rd Qu. Max.
-Inf -Inf -Inf -Inf -Inf -Inf
Fragment lengths:
int [1:233558, 1] NA NA NA NA NA NA NA NA NA NA ...
Summary of fragment lengths:
V1
Min. : 115
1st Qu.: 440
Median : 598
Mean : 609
3rd Qu.: 771
Max. :1266
NA's : 149
Fitting target prediction function to each enzyme exclusively...
Distribution of log2 signals that are finite:
Mode FALSE NA's
logical 233558 0
Distribution of units with known fragment lengths:
V1
Mode :logical
FALSE:149
TRUE :233409
NA's :0
Enzyme #1 of 1...
Distribution of units with known fragment lengths and finite signals:
Mode FALSE NA's
logical 233558 0
Distribution of units with known fragment lengths and finite signals that are exclusively on this enzyme:
Mode FALSE NA's
logical 233558 0
Error in list("process(fln, verbose = verbose)" = <environment>, "process.FragmentLengthNormalization(fln, verbose = verbose)" = <environment>, :

[2008-11-06 10:12:41] Exception: Cannot fit target function to enzyme, because there are no (finite) data points that are unique to this enzyme: 1
at throw(Exception(...))
at throw.default("Cannot fit target function to enzyme, because there are no (
at throw("Cannot fit target function to enzyme, because there are no (finite)
at getTargetFunctions.FragmentLengthNormalization(this, verbose = less(verbose
at getTargetFunctions(this, verbose = less(verbose))
at process.FragmentLengthNormalization(fln, verbose = verbose)
at process(fln, verbose = verbose)
Enzyme #1 of 1...done
Fitting target prediction function to each enzyme exclusively...done
Estimating target prediction function...done
Array #1 of 338 ('EA06028_0091-02A_Sty250_SS247725@PCRS')...done
Normalizing set for PCR fragment-length effects...done

Thanks.

Hailei

[Henrik Bengtsson]

Hi,

it looks like you haven't *fitted* the PLM, because all chip effects
(theta:s) are reported to be zero on the intensity scale or -Inf on
the log-scale, cf.

> Total thetas:
> num [1:233558] 0 0 0 0 0 0 0 0 0 0 ...
> Summary of total signals (on the intensity scale):
> Min. 1st Qu. Median Mean 3rd Qu. Max.
> 0 0 0 0 0 0
> Signals:
> num [1:233558] -Inf -Inf -Inf -Inf -Inf ...
> Summary of signals (on the log2 scale):
> Min. 1st Qu. Median Mean 3rd Qu. Max.
> -Inf -Inf -Inf -Inf -Inf -Inf

Since the fragment-length effects are modeled on the log-scale,
there are no finite data points available for fitting the model, cf.

> Fitting target prediction function to each enzyme exclusively...
> Distribution of log2 signals that are finite:
> Mode FALSE NA's
> logical 233558 0

The package detects this an throws an exception and reports the problem, cf.

> [2008-11-06 10:12:41] Exception: Cannot fit target function to enzyme,
> because there are no (finite) data points that are unique to this enzyme: 1

Please please please, when you run into problems/get errors, always
provide us with your *complete* script and your sessionInfo().

Cheers

Henrik

[hailei zhang]

I checked my script, you are right, I did not fit plm. Thanks so much.