[BioC] Memory issues with limma and ImaGen file import

[Gordon Smyth]

Dear Christian,

I have now identified the bug. It is fixed in limma 2.7.1 which 
should be available from CRAN in a couple of days.

Thanks to Peter Morrison of biodiscovery.com who enabled me to find 
the bug by sending me some example data files.

Best wishes
Gordon


>Date: Fri, 28 Apr 2006 16:21:43 +0200
>From: "Christian Spieth" <Christian.Spieth at uni-tuebingen.de>
>Subject: [BioC] Memory issues with limma and ImaGen file import
>To: <bioconductor at stat.math.ethz.ch>
>
>Hi,
>
>I am having a problem with limma 2.4.13 and ImaGene files:
>Whenever I try to read the files as described in the manual
>
>targets <- readTargets()
>files <- targets[,c("FileNameCy3","FileNameCy5")]
>RG <- read.maimages(files, source ="imagene")
>
>R aborts the operation with the following error message:
>
> > RG <- read.maimages(files,source="imagene")
>Read header information
>Fehler: kann Vektor der Gr??e 3200000 Kb nicht allozieren
>(Error: unable to allocate a vector of size 3200000 Kb)
>
>I am trying to read 10 chip files ( 5 replication with red and green).
>Each file is approx. 6MB. Is there a special format of the ImaGene files?
>My files start with
>
>
>Begin Header
>         version 5.6.1
>         Date    Mon Apr 24 20:13:17 CEST 2006
>         Image File
>X:\Lab_Hochholdinger\Individual_folders\Diana\microarray\lrt\replicate1\01Mu
>rot\0G701Murot.tif
>         Page    0
>         Page Name
>         Inverted        false
>         Begin Field Dimensions
>                 Field   Metarows        Metacols        Rows    Cols
>                 A       8       4       20      20
>                 B       4       4       20      20
>         End Field Dimensions
>         Begin Measurement parameters
>                 Segmentation Method     auto
>                 Signal Low      0.0
>                 Signal High     0.0
>                 Background Low  0.0
>                 Background High 0.0
>                 Background Buffer       3.0
>                 Background Width        5.0
>         End Measurement parameters
>         Begin Alerts
>                 Control Type    Minimum threshold       If tested
>Percentage allowed      If failed       Maximum threshold       If tested
>Percentage allowed      If failed       CV threshold    If tested       If
>failed
>         End Alerts
>         Begin Quality settings
>                 Empty Spots     true    Threshold:      2.0
>                 Poor Spots      true
>                 Begin Poor Spots Parameters
>                         Background contamination flag   true    Threshold:
>0.9995
>                         Background tested against subgrid data only     true
>                         Signal contamination flag       false   Threshold:
>0.9995
>                         Signal contamination test connected to background
>contamination threshold false
>                         Ignored percentage flag true    Threshold:      25.0
>                         Open perimeter flag     true    Threshold:      30.0
>                         Shape regularity flag   true    Threshold:      0.6
>                         Area To Perimeter Ratio flag    false   Threshold:
>0.7
>                         Offset flag     true    Threshold:      60.0
>                 End Poor Spots Parameters
>                 Negative Spots  true
>
>         End Quality settings
>End Header
>Begin Raw Data
>         Field   Meta Row        Meta Column     Row     Column  Gene ID Flag
>Signal Mean     Background Mean Signal Median   Background Median
>Signal Mode     Background Mode Signal Area     Background Area Signal Total
>Background Total        Signal Stdev    Background Stdev        Shape
>Regularity      Ignored Area    Spot Area       Ignored Median  Area To
>Perimeter       Open Perimeter  XCoord  YCoord  Diameter        Position
>offset  Offset X        Offset Y        Expected X      Expected Y      CM-X
>CM-Y    CM Offset       CM Offset-X     CM Offset-Y     Min Diam        Max
>Diam    Control Failed Control  Background contamination present
>Signal contamination present    Ignored % failed        Open perimeter
>failed  Shape regularity failed Perim-to-area failed    Offset failed
>Empty spot      Negative spot   Selected spot
>         A       1       1       1       1       HakeT2_V2-I-13-CB250166 0
>6628.6592       637.9275        6496.0  542.0   6694.4912       441.2314
>179.0   262.0   1186530.0       167137.0        2784.5909       466.856
>0.895   0.0     179.0   null    0.9763  0.0     1898.0  5548.0  16.0
>1.824   -0.7266 -1.673  1898.7266       5549.673        1898.3799
>5547.8547       1.851   -0.3467 -1.8183 14.7603 15.6522         0       0
>0       0       0       0       0       0       0       0       0
>
>......
>
>
>
>
>The error happens both under Windows (1GB RAM) and Linux (4GB RAM).
>Anyone an idea?
>
>Thanks in advance!
>
>christian
>
>
>--
>  Christian Spieth
>  Dipl.-Ing., MSc (Oxon)
>  Center for Bioinformatics (ZBIT), Univ. Tuebingen
>  NGFN - Nationales GenomForschungsNetz
>  Sand 1, D-72076 Tuebingen, Germany
>  Phone (+49/0) 7071 29 78987, Fax (+49/0) 7071 29 5091
>mailto:christian.spieth at uni-tuebingen.de
>  http://www-ra.informatik.uni-tuebingen.de
>
>  PGP fingerprint =  8A AC FD DA 57 A3 15 67  23 16 15 0A BD 04 AC A7  MD5
>  Fingerprint =  22a9627dcc5302371de7764b40c2be6d