Is it sensable to use p-value detection in illumina bead array as present, absent or marginal call
3
1
Entering edit mode
@agaz-hussain-wani-7620
Last seen 6.0 years ago
India

I dealt with some Affymetrix data , a part of which is pasted below

ID_REF
VALUE
ABS_CALL
DETECTION P-VALUE
10071_s_at
 3473.6
 P
         0.000219
1053_at 
 643.2
 P
         0.000673
117_at
 564
 P
         0.000322
1255_g_at
 9.4
 A
         0.602006
1294_at
 845.6
 P
         0.000468
1320_at
 94.3
 A
         0.204022
1405_i_at
 6546.2
 M
         0.0631
14312_at
 54.1
 P
         0.003067
1438_at
 461.3
 P
         0.000562

 

Where i easily can decide the calls either Present, Absent or Marginal. 

I have some illumina bead array data also, shown below

ID
ILMN_1681101
Pvalue    Intenstiy
0.27403

6.966361247
ILMN_2094942
0.18961
7.00337736
ILMN_1703142
0
7.600470477
ILMN_2271336
0.37662
6.935459748
ILMN_2337789
0.08312
7.064877464
ILMN_1669592
0.00519
7.24596858
ILMN_1735038
0.05325
7.089582893

Can i use pvalue here to make Present, Absent or Marginal call same as Affymetrix data. Thanks

illumina pvalue • 2.8k views
ADD COMMENT
2
Entering edit mode
@gordon-smyth
Last seen 6 hours ago
WEHI, Melbourne, Australia

Yes. See the Illumina BeadChip case study in Section 17.3 of the edgeR User's Guide, which reads the detection p-values and uses them to filter probes.

You might also find this article interesting: http://www.ncbi.nlm.nih.gov/pubmed/20056656

ADD COMMENT
1
Entering edit mode
svlachavas ▴ 830
@svlachavas-7225
Last seen 5 months ago
Germany/Heidelberg/German Cancer Resear…

 Dear hussainaaghaz,

yes it is sensible and useful to use the DetectionValue in Illumina to determine whether or not a probe is detected above a threshold level in each of the samples in your experiment. Thus, if you use the default detection p-value threshold( < 0.01), you can use the below naive functions to remove a probe from all the samples if is not detected on any of your total number of arrays. So a probe that is detected on at least one sample remains:

present_probes <- detectionCall(lumi_data) # lumi_data your raw data prior normalization

selected_probes <- exprs(norm_data)[present_probes >0, ]

and then you can see how many probes have remained and check your intensity distribution with some plots like histogram or densityPlot

 

ADD COMMENT
0
Entering edit mode

Additionaly if you have preprocessed your data with limma,

you can use 

expressed_probes <- rowSums(norm.data$other$Detection < 0.01) >=N #  N=Number of samples you want to be present and norm.data your normalized set class:"EList"
filtered <- norm.data[expressed,]

ADD REPLY
0
Entering edit mode

Dear Svlachavas, thanks for your comments.

ADD REPLY
0
Entering edit mode
@agaz-hussain-wani-7620
Last seen 6.0 years ago
India

Thanks dear Smyth for help and pointing to reference.

ADD COMMENT

Login before adding your answer.

Traffic: 989 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6