Results only being calculated randomly

<p> We had an issue recently where the custom field reporting results is randomly giving all zeros for some samples in the run and giving the correct %area result for others.  The custom field I am using is:</p><p></p><p>((Area/(D..(XYZ[Area])))*10)+0</p><p></p><p>I have set search order to results set only (not sure if that could be the issue?), sample type is unknowns and peak type is all.  I have ticked the missing peak.  Its a peak, real , calculated field.  I have added the +0 at the end to report as 0.00000 if nothing is detected.  Basically, the calculation is the area of any peak in the injection divided by the area of the peak XYZ in the injection with label D in the sample set multiplied by 10.</p><p></p><p>When we process, only some of the samples have results for this field, even though the same area value from the channel labelled D is used.  We are batch processing.  If we use the existing integration option, now some results that showed 0.00000 suddenly give results.  it appears very random and is affecting one sample set at present.  Another sample set processed later worked fine.</p><p></p><p>Has anyone seen anything like this before?</p><p></p><p>Thanks</p>

Comments

  • Hi,

    Haven't had this issue before, but we do a lot of work with custom fields.

    There are a few things which I suspect you will have checked already, but which could cause this kind of issue.

    Do all the samples have an Area for the components in the table? The calculation will still fault if the peak is missing as it won't have an Area value to reference.

    Similarly, does the last result processed with label D (not necessarily the last sample injected with label D) have a peak area for XYZ?

    The processing order isn't necessarily the same as the injection order, so it might be worth using processing commands in the sample set to ensure that the sample or samples with label D are definitely processed before the samples with which they are compared.

    When a result set gave different results when re-processed using existing integration, am I right in thinking that some of the samples had been re-integrated?

    There might also be a problem if the sample labelled D is a Standard. Does the calculation produce the same results if the Sample Type is set to All?

    If you want to return 0 for "no area" and assuming there will always be a XYZ[Area] in sample D, does this field work instead?

    10*REPLACE(Area,0)/D..(XYZ[Area])

    Hope this helps, or at least clarifies the issue for others.

    Regards,

    Kenny.

  • Hi Kenny,

    Thanks for the reply.

    Yes all of the components have an area in the results table but the custom field I require still shows 0.000000.  There is just the one injection for sample with label D.  It is basically a dilute solution (1/10) of any sample and the main peak area from that injection is used to calculate the impurity content for all samples in the run.  Its an old method from the days when chromatograms used to top off for the main peak.

    Since posting this, I have realised that the channels were not processed in the order they appear in the sample set.  So I have the following:

    Dilute Sample (Label D)

    Sample 1

    Sample 2

    Sample 3

    Sample 4

    Empower processed as follows (as an example):

    Sample 3

    Sample 4

    Dilute Sample (Label D)

    Sample 1

    Sample 2

    So only samples 1 and 2 have results calculated as samples 3 and 4 do not have the required input value from the Dilute sample.  Is there any reason this would happen on Empower?  I have a quantitate line and summarize custom fields line at the end of the sample set.  So I thought it would process in order of injections and summarize any intersample custom fields at the end?  Should I use other functions or labels?  I am not using U101, U102 etc labels here.

  • It could be worthwhile giving all the samples a label just to be absolutely sure about the processing order:

    LabelSampleProcessingProcessing CommandLabel Reference
    DDilute SampleDon't Process or Report
    U01Sample 1Don't Process or Report
    U02Sample 2Don't Process or Report
    U03Sample 3Don't Process or Report
    U04Sample 4Don't Process or Report
    NormalQuantitateD
    NormalQuantitateU*
    NormalSummarize Custom Fields

    I can only guess as to why the processing order would be different from the injection order, but I think at least some of the processing occurs in parallel. I've seen the odd injection process "out of order", so I tend to ensure that my reports don't depend on the Result ID or Processing Date. It might happen if it takes longer to process the dilute sample than it does to process the others. For example, if the dilute sample has a longer run-time, has more peaks, uses a different acquisition data rate or uses a different processing method it could finish after the earlier samples...

    I think the processing commands should sort this out though.

    Best of luck!

    Kenny.

  • Hi Kenny,

    The labels worked a treat! Thanks for that.  I am still unclear why it processes channels in incorrect order as the dilute sample only has one peak to be integrated so it should not take that long.  At least the labels will ensure it works 100% of the time.

    Thanks