[BioC] altcdfenvs

[Laurent Gautier]

Hee Siew Wan wrote:
> Thanks Laurent for the tip but I encountered other problems if I
> create enough identifiers. When I created one unique identifier for
> each probe pair I want to be inside the new cdf, then I would get
> 99112 probe *sets* because

You need one identifier per probe set, not for each probe pair !!!

A probe set is made of probe pairs, and a probe pair is made of two 
probes (the PM and its corresponding MM).

Did you actually follow the main vignette for the package ?
The example uses objects of small size to let one experiment and understand.
If you run it and take the objects 'm', and 'ids':

 > length(m$match)
[1] 3
 > length(ids)
[1] 3
 > m$match
[[1]]
  [1] 77288 77289 77290 77291 77292 77293 77294 77295 77296 77297 77298 
77299
[13] 77300 77301 77302 77303 77304 77305 77306 77307 77308 77309

[[2]]
  [1] 100047 100048 100049 100050 100051 100052 100053 100054 100055 100056
[11] 100057 100058 100059 100060 100061 100062 100063 100064 100065 100066
[21] 100067 100068 122174

[[3]]
  [1]  34732  34733  34734  34735  34736  34737  34738  34739  34740  34741
[11]  34742 122168 122169 122170 122171 122172 122174 122175 122176 122177
[21] 122178
 > alt.cdf <- buildCdfEnv.matchprobes(m, ids,
                                    nrow.chip = 712, ncol.chip = 712,
                                    chiptype = "HG-U133A",
                                    probes.pack = "hgu133aprobe")
 > alt.cdf
Instance of class CdfEnvAffy:
  name     : HG-U133A
  chip-type: HG-U133A
  size     : 712 x 712
  3 probe set(s) defined.


> 
>> length(unique(ind))
> 
> [1] 99112
> 
> This isn't exactly what I have in mind. If say, I want to have 4
> probe pairs (nearest to the 5-prime end) from each set, how can I
> proceed to create this new cdf?

To be 'near the 5-prime end', you will need to a) have reference 
sequences and match the probes against them, or b) get that information 
from somewhere (NetAffx may be). If I were you, I would not take risk 
with the ordering in the probe sequence table.

> What I realised from what I've done below is that I will get one
> probe pair that's furthest from 5-prime end for each set because the
> furthest pair is at the *bottom* of the probe set. The probe table is
> arranged in increasing order and so it seems to me that it updates
> itself and did not keep the earlier ones.

I am sorry, but I am really confused...
You may also have to explain more about what you have in mind (off-list, 
if you wish) if you want someone to help.



L.

> Please advice and thanks for the help.
> 
> Cheers sw
>