Microarray processed data: Cut off

Question

[Payton Yau]

Hi all,

I am new in here and have some questions about array analysis.

1. How can I make a cut-off in a processed micro-array?
2. If I found a specific cut-off in one specific gene for analysis, do I need to use the same cut off for the whole data set, if I want to compare with the other genes?

Thanks

[Ahill]

Regarding question (1) - take a look at the "similar posts" listed on the right hand side of the page here - there are good answers on those threads about possible data filters, for example Microarray Data: Intensity Cutoff

Regarding question (2) - possibly, but it depends on what type of cut-off you are thinking of - an intensity cutoff? An Affymetrix A/P call cutoff? Or a statistic-based cutoff (like an adjusted p-value from a T-test)?

The answer depends on what type of microarray you are using. If you provide that information, others could give you more detailed advice.

[Payton Yau]

Thank you for your help. I will look further about the old posts.

I am analysing the normalised data from whole-Genome DASL HT Assay, illumina and I think I am using a statistic-based cut-off. The array value ranging from 6.6 - 15.

For example, the cut-off from a specific gene A in 9 has statistically significant and gene B in 10 has statistically significant. Do I need to keep the cut-off 9 or 10 for further gene-gene inter-reaction analysis, or other value cut-off (present / absent) I should use?

[TriS]

what are the numbers you mentioned (6.6-15 range and 9, 10)? can you paste some of those data here (10-20 lines) so that we can have a look? I have a feeling that as Ahill mentioned is a p.value based cutoff (or adjusted p.value) but as the post is written it's pretty hard to say.

[Payton Yau]

For example, gene A cut-off is 10.7537. I want to use gene A to compare with gene B,C D... in the array data,hence, should I use the gene A cut-off for all the data set, or using other cut-off (such as mid-point or average from the whole data), for chi-square calculation?

I don't have any experience to handle array data, hope someone can help me, thanks.

A           B           C           D           E           F           H
6.8189      13.0030     8.6803      10.8104     9.5281      11.6192     7.1449
7.9735      13.3488     9.6836      11.0574     7.8270      11.0267     6.9630
7.9847      13.1529     8.9669      11.0566     10.5105     10.1033     6.8814
8.6803      13.2476     7.9847      11.2119     10.4203     11.3937     7.0782
8.8414      7.0880      13.1466     10.9931     6.7923      11.2176     6.7820
9.0455      9.6054      13.2458     10.9381     7.4185      11.4213     7.0444
9.7937      8.9361      13.3972     6.9561      10.3311     10.2571     6.9319
10.3126     7.3138      13.3550     7.1564      7.0120      12.1360     10.9988
10.7537     7.6823      13.1720     6.7970      8.7083      11.8906     10.9523 #the first number (10.7537 is the cutoff)
10.7799     7.3751      13.3529     6.8419      8.4394      11.1961     11.0398
10.8189     7.9735      13.4438     7.7458      9.0455      11.6035     11.0252
11.0252     7.8044      13.5171     11.1341     9.0650      11.3855     11.0194
11.7450     7.7662      13.4604     10.9341     7.0559      12.4262     8.7895
12.7937     7.9735      13.3368     11.1066     9.9627      11.9987     8.4371
13.5306     7.7620      13.2156     10.7855     6.9055      12.1645     7.0442
13.5854     7.8120      13.3886     10.8021     6.9668      11.4794     6.9840
13.6758     7.4405      13.3740     10.6626     7.2063      12.0295     8.3033
14.8189     9.1630      13.5982     9.0114      7.1289      11.7442     6.9820
14.9548     7.7369      13.5918     10.6889     8.7099      11.5042     7.0968

[TriS]

OK, I'm a bit confused, why do you want to do chi square? You need a contingency table for chi-square, I'm not sure that's what you want.

1. How did you get the 10.7537 as cutoff?
2. Is your question: "is gene A expression different from B, C, etc etc?
3. I assume that in the matrix you posted the columns are genes and rows are samples?