5623 |
06 May 11 |
nicklas |
1 |
/** |
5623 |
06 May 11 |
nicklas |
$Id$ |
5623 |
06 May 11 |
nicklas |
3 |
|
5623 |
06 May 11 |
nicklas |
Copyright (C) 2011 Nicklas Nordborg |
5623 |
06 May 11 |
nicklas |
5 |
|
5623 |
06 May 11 |
nicklas |
This file is part of BASE - BioArray Software Environment. |
5623 |
06 May 11 |
nicklas |
Available at http://base.thep.lu.se/ |
5623 |
06 May 11 |
nicklas |
8 |
|
5623 |
06 May 11 |
nicklas |
BASE is free software; you can redistribute it and/or |
5623 |
06 May 11 |
nicklas |
modify it under the terms of the GNU General Public License |
5623 |
06 May 11 |
nicklas |
as published by the Free Software Foundation; either version 3 |
5623 |
06 May 11 |
nicklas |
of the License, or (at your option) any later version. |
5623 |
06 May 11 |
nicklas |
13 |
|
5623 |
06 May 11 |
nicklas |
BASE is distributed in the hope that it will be useful, |
5623 |
06 May 11 |
nicklas |
but WITHOUT ANY WARRANTY; without even the implied warranty of |
5623 |
06 May 11 |
nicklas |
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
5623 |
06 May 11 |
nicklas |
GNU General Public License for more details. |
5623 |
06 May 11 |
nicklas |
18 |
|
5623 |
06 May 11 |
nicklas |
You should have received a copy of the GNU General Public License |
5623 |
06 May 11 |
nicklas |
along with BASE. If not, see <http://www.gnu.org/licenses/>. |
5623 |
06 May 11 |
nicklas |
21 |
*/ |
5623 |
06 May 11 |
nicklas |
22 |
package net.sf.basedb.util.affymetrix; |
5623 |
06 May 11 |
nicklas |
23 |
|
5623 |
06 May 11 |
nicklas |
24 |
import affymetrix.fusion.cdf.FusionCDFData; |
5623 |
06 May 11 |
nicklas |
25 |
import affymetrix.fusion.cel.FusionCELData; |
5623 |
06 May 11 |
nicklas |
26 |
import net.sf.basedb.core.File; |
5623 |
06 May 11 |
nicklas |
27 |
import net.sf.basedb.core.InvalidDataException; |
5623 |
06 May 11 |
nicklas |
28 |
import net.sf.basedb.core.InvalidUseOfNullException; |
5623 |
06 May 11 |
nicklas |
29 |
import net.sf.basedb.core.ItemNotFoundException; |
5623 |
06 May 11 |
nicklas |
30 |
import net.sf.basedb.core.RawBioAssay; |
5623 |
06 May 11 |
nicklas |
31 |
import net.sf.basedb.core.StringUtil; |
5623 |
06 May 11 |
nicklas |
32 |
|
5623 |
06 May 11 |
nicklas |
33 |
/** |
5623 |
06 May 11 |
nicklas |
Helper class for working with Affymetrix CEL files. |
5623 |
06 May 11 |
nicklas |
35 |
|
5623 |
06 May 11 |
nicklas |
@author Nicklas |
5623 |
06 May 11 |
nicklas |
@since 3.0 |
5623 |
06 May 11 |
nicklas |
@base.modified $Date$ |
5623 |
06 May 11 |
nicklas |
39 |
*/ |
5623 |
06 May 11 |
nicklas |
40 |
public class CelValidator |
5623 |
06 May 11 |
nicklas |
41 |
{ |
5623 |
06 May 11 |
nicklas |
42 |
|
5623 |
06 May 11 |
nicklas |
43 |
/** |
5623 |
06 May 11 |
nicklas |
Create a new instance. |
5623 |
06 May 11 |
nicklas |
45 |
*/ |
5623 |
06 May 11 |
nicklas |
46 |
public CelValidator() |
5623 |
06 May 11 |
nicklas |
47 |
{} |
5623 |
06 May 11 |
nicklas |
48 |
|
5623 |
06 May 11 |
nicklas |
49 |
private void setHeader(RawBioAssay rba, String name, String value) |
5623 |
06 May 11 |
nicklas |
50 |
{ |
5623 |
06 May 11 |
nicklas |
51 |
if (name == null || value == null) return; |
5623 |
06 May 11 |
nicklas |
52 |
value = StringUtil.trimString(value, RawBioAssay.MAX_HEADER_VALUE_LENGTH); |
5623 |
06 May 11 |
nicklas |
53 |
rba.setHeader(name, value); |
5623 |
06 May 11 |
nicklas |
54 |
} |
5623 |
06 May 11 |
nicklas |
55 |
|
5623 |
06 May 11 |
nicklas |
56 |
/** |
5623 |
06 May 11 |
nicklas |
Load a CEL file using the Affymetric Fusion SDK. This method |
5623 |
06 May 11 |
nicklas |
checks that the file exists, and read all headers. |
5623 |
06 May 11 |
nicklas |
59 |
|
5623 |
06 May 11 |
nicklas |
@param celFile The file to load |
5623 |
06 May 11 |
nicklas |
@return A FusionCELData object representing the CEL file |
5623 |
06 May 11 |
nicklas |
@throws ItemNotFoundException If the actual file is not on the server |
5623 |
06 May 11 |
nicklas |
@throws InvalidDataException If the file is not a CEL file |
5623 |
06 May 11 |
nicklas |
64 |
*/ |
5623 |
06 May 11 |
nicklas |
65 |
public FusionCELData loadCelFile(File celFile) |
5623 |
06 May 11 |
nicklas |
66 |
{ |
5623 |
06 May 11 |
nicklas |
67 |
if (celFile == null) throw new InvalidUseOfNullException("celFile"); |
5623 |
06 May 11 |
nicklas |
68 |
if (!celFile.getLocation().isDownloadable()) |
5623 |
06 May 11 |
nicklas |
69 |
{ |
5623 |
06 May 11 |
nicklas |
70 |
throw new ItemNotFoundException("Can't download file data for file '" + |
5623 |
06 May 11 |
nicklas |
71 |
celFile.getName() + "'; location=" + celFile.getLocation()); |
5623 |
06 May 11 |
nicklas |
72 |
} |
5623 |
06 May 11 |
nicklas |
73 |
FusionCELData cel = new FusionCELData(); |
5623 |
06 May 11 |
nicklas |
74 |
cel.setFileName(celFile.getName()); |
5623 |
06 May 11 |
nicklas |
75 |
cel.setInputStream(celFile.getDownloadStream(0)); |
5623 |
06 May 11 |
nicklas |
76 |
if (!cel.readHeader()) |
5623 |
06 May 11 |
nicklas |
77 |
{ |
5623 |
06 May 11 |
nicklas |
78 |
throw new InvalidDataException("Could not read CEL file '" + |
5623 |
06 May 11 |
nicklas |
79 |
celFile.getPath()+"': " + cel.getError()); |
5623 |
06 May 11 |
nicklas |
80 |
} |
5623 |
06 May 11 |
nicklas |
81 |
return cel; |
5623 |
06 May 11 |
nicklas |
82 |
} |
5623 |
06 May 11 |
nicklas |
83 |
|
5623 |
06 May 11 |
nicklas |
84 |
/** |
5623 |
06 May 11 |
nicklas |
Copy metadata from the CEL file to the raw bioassay. |
5623 |
06 May 11 |
nicklas |
This will set the mumber of spots and headers from the |
5623 |
06 May 11 |
nicklas |
file. |
5623 |
06 May 11 |
nicklas |
88 |
|
5623 |
06 May 11 |
nicklas |
@param cel The CEL file |
5623 |
06 May 11 |
nicklas |
@param rba The raw bioassay |
5623 |
06 May 11 |
nicklas |
91 |
*/ |
5623 |
06 May 11 |
nicklas |
92 |
public void copyMetadata(FusionCELData cel, RawBioAssay rba) |
5623 |
06 May 11 |
nicklas |
93 |
{ |
5623 |
06 May 11 |
nicklas |
94 |
rba.setNumFileSpots(cel.getCells()); |
5623 |
06 May 11 |
nicklas |
95 |
|
5623 |
06 May 11 |
nicklas |
96 |
setHeader(rba, "Algorithm", cel.getAlg()); |
5623 |
06 May 11 |
nicklas |
97 |
for (int i = 0; i < cel.getNumberAlgorithmParameters(); ++i) |
5623 |
06 May 11 |
nicklas |
98 |
{ |
5623 |
06 May 11 |
nicklas |
99 |
String name = cel.getAlgorithmParameterTag(i); |
5623 |
06 May 11 |
nicklas |
100 |
setHeader(rba, "Algorithm parameter: " + name, |
5623 |
06 May 11 |
nicklas |
101 |
cel.getAlgorithmParameter(name)); |
5623 |
06 May 11 |
nicklas |
102 |
} |
5623 |
06 May 11 |
nicklas |
103 |
setHeader(rba, "Dat header", cel.getDatHeader()); |
5623 |
06 May 11 |
nicklas |
104 |
setHeader(rba, "Chip type", cel.getChipType()); |
5623 |
06 May 11 |
nicklas |
105 |
} |
5623 |
06 May 11 |
nicklas |
106 |
|
5623 |
06 May 11 |
nicklas |
107 |
/** |
5623 |
06 May 11 |
nicklas |
Reset metadata on the given raw bioassay. This set the |
5623 |
06 May 11 |
nicklas |
number of spots to 0 and remove all headers. |
5623 |
06 May 11 |
nicklas |
@param rba The raw bioassay |
5623 |
06 May 11 |
nicklas |
111 |
*/ |
5623 |
06 May 11 |
nicklas |
112 |
public void resetMetadata(RawBioAssay rba) |
5623 |
06 May 11 |
nicklas |
113 |
{ |
5623 |
06 May 11 |
nicklas |
114 |
rba.setNumFileSpots(0); |
5623 |
06 May 11 |
nicklas |
115 |
rba.removeHeaders(); |
5623 |
06 May 11 |
nicklas |
116 |
} |
5623 |
06 May 11 |
nicklas |
117 |
|
5623 |
06 May 11 |
nicklas |
118 |
/** |
5623 |
06 May 11 |
nicklas |
Check if the loaded CEL and CDF files matches. The chip type is |
5623 |
06 May 11 |
nicklas |
verified. Unfortunately the chip type is not stored inside the CDF |
5623 |
06 May 11 |
nicklas |
file, but has to be given as an extra parameter. Typically, this |
5623 |
06 May 11 |
nicklas |
should be the name of the file. This method also verify that the number |
5623 |
06 May 11 |
nicklas |
rows and columns in the two files matches. |
5623 |
06 May 11 |
nicklas |
124 |
|
5623 |
06 May 11 |
nicklas |
@param cel The CEL fila data |
5623 |
06 May 11 |
nicklas |
@param cdf The CDF file data |
5623 |
06 May 11 |
nicklas |
@param cdfChipType The CDF chip type |
5623 |
06 May 11 |
nicklas |
128 |
*/ |
5623 |
06 May 11 |
nicklas |
129 |
public void validateCelAndCdf(FusionCELData cel, FusionCDFData cdf, String cdfChipType) |
5623 |
06 May 11 |
nicklas |
130 |
throws InvalidDataException |
5623 |
06 May 11 |
nicklas |
131 |
{ |
5623 |
06 May 11 |
nicklas |
132 |
String celChipType = cel.getChipType(); |
5623 |
06 May 11 |
nicklas |
133 |
if (!cdfChipType.startsWith(celChipType)) |
5623 |
06 May 11 |
nicklas |
134 |
{ |
5623 |
06 May 11 |
nicklas |
135 |
throw new InvalidDataException("CEL chip type (" + celChipType + |
5623 |
06 May 11 |
nicklas |
136 |
") doesn't match CDF chip type (" + cdfChipType + ")"); |
5623 |
06 May 11 |
nicklas |
137 |
} |
5623 |
06 May 11 |
nicklas |
138 |
|
5623 |
06 May 11 |
nicklas |
139 |
if (cel.getRows() != cdf.getHeader().getRows() || cel.getCols() != cdf.getHeader().getRows()) |
5623 |
06 May 11 |
nicklas |
140 |
{ |
5623 |
06 May 11 |
nicklas |
141 |
throw new InvalidDataException("CEL size (rows=" + cel.getRows() + ", cols=" + cel.getCols() + |
5623 |
06 May 11 |
nicklas |
142 |
") doesn't match CDF size (rows=" + |
5623 |
06 May 11 |
nicklas |
143 |
cdf.getHeader().getRows() + ", cols=" + cdf.getHeader().getCols()+")"); |
5623 |
06 May 11 |
nicklas |
144 |
} |
5623 |
06 May 11 |
nicklas |
145 |
} |
5623 |
06 May 11 |
nicklas |
146 |
|
5623 |
06 May 11 |
nicklas |
147 |
|
5623 |
06 May 11 |
nicklas |
148 |
|
5623 |
06 May 11 |
nicklas |
149 |
} |