If there's a related error in the log you should post it.

This is an unusual error… is this on a released build or a custom, dev build?

The error is saying that this particular run (with RunId = 551) was not found in the database. But if you were able to select it in the runs list then it must exist. So you'll have to do some debugging to figure out what's going wrong. First, look in your MS2Runs table to see if the run exists. If so, you'll need to step through code like MS2Manager.getRuns(). Perhaps you're storing runs with a new type… and there's no matching MS2Run subclass to handle it?

Adam

Does the error happen if you do not select any specific view?
*Chee-Hong

Hi,

Well the version that I am running is 1.5 with Mascot support. The files that I am looking at were created from mzXML files using X!Tandem. There are 86 files but 6 or so did not process properly. If I load twenty of the files then I can export the data to Excel whether I apply a custom view or not. I if I try to load more than that then I receive an error, Even if the data are fltered with a custom view, I just checked now and I can export all 80 files to tsv format so The problem must be the row limit in excel. Perhaps then I need more stringent filtering. I have attached some log files and generated error files from he browser.

Thanks,

Bill

Excel export is certainly limited to 65,535 rows, but exceeding that shouldn't generate any errors… the output is just truncated. Filtering vs. not shouldn't make a diffence either (obviously you get to the 64K limit sooner without a filter, but still not an error).

The error you initially reported is generated when the selected runs are loaded from the MS2Runs table… this is before any exporting or filtering takes place. It seems to have failed to load this particular run. Perhaps this run is one that did not process properly? That might explain the error. Either way, at least put a breakpoint on the line of code that generates this error and see what's going on.

Much of this code is changing in 1.6 to provide better error recovery/reporting and allow arbitrarily large lists of runs for export & compare.

Adam

Here is an update:

The first reported error message 'Run 551: Not Found' is run that failed to be loaded because the peptideProphet computation of <distribution_point> generate some special floating numbers (e.g. NaN, infinite, indefinite). The output seems to be different in C for Windows and Linux vs that of Java. Specifically, "-1.#J" is found in the pepXML file and it is not parsable by Java, thus the MS2 load failed.

The export is however failing on:
java.lang.ArrayIndexOutOfBoundsException: 441
    at jxl.biff.IndexMapping.getNewIndex(IndexMapping.java:68)
    at jxl.biff.FormattingRecords.rationalize(FormattingRecords.java:385)
    at jxl.write.biff.WritableWorkbookImpl.rationalize(WritableWorkbookImpl.java:878)
    at jxl.write.biff.WritableWorkbookImpl.write(WritableWorkbookImpl.java:599)
    at org.fhcrc.cpas.data.ExcelWriter.closeWorkbook(ExcelWriter.java:403)
    at MS2.MS2Controller.exportRuns(MS2Controller.java:2579)

This does not happen for run with 2, 10, 15, and 20 selections. For 30, 41, 47, and 66 selection, the above exection will happen.

It does not seems like the row limit has been exceeded yet as "jxl.write.biff.RowsExceededException: The maximum number of rows permitted on a worksheet been exceeded" is not found in the log.

However, the formatting objects are not static and the write order of the rows and columns are correct that the usual causes of this exception are not responsible. I tried to reproduce this by exporting 100 runs at once, but to no avail. Any other suggestion?

Chee-Hong

And I wanted to emphasize that the same list of files (80 of them)that gives the error when I try to export to Excel (either of the two choices) is exported when I select TSV format.

Thanks

Bill

Not to point fingers, but my best guess is that this is a problem with JXL. A couple threads on the Java forums may be talking about this problem (or not):

• http://forum.java.sun.com/thread.jspa?threadID=688027
• http://forum.java.sun.com/thread.jspa?threadID=634804

The first one suggests upgrading to version 2.5.9 or later (CPAS is currently shipping 2.5.7), though there are no other details and nothing obviously causing your problem in the JXL change log between these two versions.

Since I can't reproduce the problem, it would be great if you could download the 2.6 version of jxl.jar (http://www.andykhan.com/jexcelapi/download.html), replace yours, and try to reproduce. If it fixes the problem (and doesn't cause any new problems) I will include the latest version in the next release.

If this doesn't solve the problem we'll probably have to debug JXL to figure it out.

Okay, so I tried to get version 2.5.9 and also 2.6.0 and when I tried to unzip it (WindowsXP-Pro) I receive an error: Error reading header after processing 0 entries. I am willing to try this but do not know what to do next.

Thanks,

Bill

I've extracted the 2.6 jxl.jar file (using WinZip) and attached it below. Hopefully that will work for you.

Thanks for the jar fle. Unhappily I received the same error. Attached is the error message that I received. Maybe I should wait for 1.6.

Thanks,

Bill

I'm changing some of the export code in 1.6, but I doubt these changes will affect this problem. You could try syncing up to the trunk and trying this on a current build.

Have you been able to narrow down the repro case at all? Is it one particular run that causes the exception? Or only a specific set of runs? Any different sets that result in the problem?

I may have to get your data to try to reproduce & debug the problem here.

Hi Adam, Sorry for the delay on this. I have only a couple of sets of data in CPAS but they all exhibit the same problem: above about 30 selections the problem occurs. I will be glad to provide data that demonstrates this. Please let me know where to post it. If you wish I can post it on our public ftp site.

Thanks,

Bill

Hi Bill,

We've now posted CPAS 1.6… it'd be great if you could try this on the new build. I suspect the problem will still appear, but it's worth checking.

I have tracked down one problem with exporting multiple runs to Excel. Turns out that we don't protect against exceeding maximum rows when we're rendering the inter-run header. I'm reworking the ExcelWriter for 1.7 to be more robust in this case. If this is the only problem, then you should be able to export using the "Excel with minimal header text" format without ever hitting an exception.

Let me know if you are able to confirm this.

Thanks, Adam

Hi Adam,

Well I have had some success with this. If I try to upload my data to excel without applying a filter then neither one of the Excel choices works and I get "This page cannot be displayed: HTTP 500 - Internal server error Internet Explorer."

But if I Apply a filter

Peptide Filter: (Hyper >= 20) AND ((Peptide SIMILAR TO 'KR.^P%' OR Peptide SIMILAR TO '-.%') AND (Peptide SIMILAR TO '%KR.^P' OR Peptide SIMILAR TO '%.-')) Peptide Sort: Protein Group Filter: (UniquePeptidesCount >= 2) AND (GroupProbability >= 0.9) Protein Group Sort:

then the Spreadsheet opens.

This is with about 80 samples and while running version 1.6.

Thanks,

Bill