Affymetrix probe id

Question

[rv]

Hello everyone,

I know that affymetrix probeset ids are portable across platforms: the same probeset id 1090_f_at always refers the same gene (probe set). Is it the same for probe ids?

Executing, as discussed in the "affy" package vignette, the code:

> pm(Dilution, "1090_f_at")

on the chip HG_U95Av2, for the probeset 1090_f_at are returned its pm probes 1, 2, .. 16

                            20A 20B 10A 10B
1090_f_at1          115.0 74.0 94.0 61.0
1090_f_at2          129.3 80.3 108.0 70.3
..
1090_f_at16         262.0 157.0 194.8 131.5

Executing the same code on a different chip containing, of course, the same probeset 1090_f_at, are its pm probes returned in the same order, so that their ids

1090_f_at1,
1090_f_at2,
..
1090_f_at16

are consistent across chips?

I suppose so. Could anyone confirm it? How could I check it?

Thank you

Answer 1

[Jason H]

It may depend on the source of the file you are using to obtain probe ids. Probe IDs in files from Affymetrix are integers and are associated with a specific feature (x, y position) on the microarray. For clarity this is different than a probeset id. A probeset id refers to a collection of probes (you seem to know this difference but I see people use the terms probe and probeset interchangeably in casual conversation which leads to confusion...)

Probe IDs for Affymetrix arrays (in files released from Affymetrix, this may differ depending on the creator of your analysis package) are often calculated based on a formula and convertible from the probe id to an x y coordinate on the microarray.

Therefore the only reason a probe_id should be the same between two different arrays is if the probe is the same height and width, and the probe is at the same x, y coordinate on the microarray. (This is rare).

You can download lots of information from the Affymetrix website. If you have any further questions just ask. I'm pretty familiar with it.

[rv]

Thank you very much @Jason H,

I am working with two very similar chips hgu133a and ht-hgu133a in R by the "affy" package. Do you think in this case I could assume the consistency of the probe ids as they are returned by the pm() method from previous example? I know, I should ask the author of the "affy" package to be sure: however, in case the x,y coordinates of each probe were the same between the two so strictly related platforms I would assume the consistency.

Is there a way to map the probe id to the oligonucleotide sequence? If so, I could map the same probe from one chip to the other even in case the probe ids returned by the pm() method were inconsistent.

I could also leave the "affy" package and directly elaborate affymetrix files if it were possible, by them, to map the same probe from one chip to the other.

[Jason H]

I think you are looking for these (Note - this assumes you are using HG-U133A and not HG-U133A_2. I don't think HG-U133A_2 has an HT version but I'm not certain)

• http://www.affymetrix.com/analysis/downloads/data/HG-U133A.probe_tab.zip
• http://www.affymetrix.com/analysis/downloads/data/HT_HG-U133A.probe_tab.zip

The probe_id and sequence is the probe_id that matches the CDF file distributed by Affymetrix for these arrays. Different packages, such as the one you are using, may assign different probe_ids.

Hope this is somehow helpful.

[rv]

Yes, it is.

In the tables, the "Probe Interrogation Position" is the position of the 13º (central) base of the probe nucleotide sequence in the gene sequence. I guess that probe ids returned by the "affy" package 1, 2, .. 16 are likely to be sorted by such a position and then they are likely to be consistent across platforms. However, I have directly asked the maintainer (prof. Irizarry) of the "affy" package: I will post his answer if I will receive.

Thank you @Jason H