convertCel(..., newChipType) fail for Calvin/v1 CEL files

[HenrikBengtsson]

convertCel(..., newChipType) updated 'chiptype' and 'header' fields in the CEL header, but not the datheader field. For example, a CEL header may look like:

Browse[2]> str(hdr)
List of 14
 $ filename      : chr "C:/Users/hb/R/win-library/3.3/AffymetrixDataTestFiles/ra
wData/FusionSDK_Test3/Test3/2.Calvin/Test3-1-121502.CEL"
 $ version       : int 1
 $ cols          : int 126
 $ rows          : int 126
 $ total         : int 15876
 $ algorithm     : chr "Percentile"
 $ parameters    : chr "Percentile:75;CellMargin:2;OutlierHigh:1.500;OutlierLow:
1.004;GridULX:154.000000;GridULY:164.000000;GridURX:995.000000;GridURY:"| __trun
cated__
 $ chiptype      : chr "Test3"
 $ header        : chr ""
 $ datheader     : chr "[12..40151]  Fetal 3:CLS=1167 RWS=1167 XIN=3  YIN=3  VE=
17        2.0 08/16/01 17:28:31    \024  \024 Test3.1sq \024  \024  \02"| __trun
cated__
 $ librarypackage: chr ""
 $ cellmargin    : int 2
 $ noutliers     : int 5
 $ nmasked       : int 0

Note how the chip type is encoded also in the datheader fields. In other words, depending on CEL file/header, the output may or may not get the new chip type set.

[HenrikBengtsson]

It looks like .getCelHeaderV4() is the bad guy. For Calvin/v1 CEL files, .getCelHeaderV4(header) will re-read the CEL header, use it to override the header already set.