Custom Fields

Hi ,

What is the best process to test custom fields ?

I Have test example sample data and quantifying assay and rel subs OK by calibrate quantitate but no custom field information 

Do you need to process using specific processing/method set before seeing CF working or do CF's require to be throughout sample set depending on labelling ?

Can you see results if you have multiple CF (intersample) or duplicate injections summary with 'fingers crossed' and click process  ?


Thanks

Answers

  • It depends on the type of CF you create. And the search order plays a role when testing intersample/intersample summary CFs. for example, if you are creating a CF like Area*CConst1, I would populate CConst1 for at least 3 peaks, process the sample set and compare the full values of Area*CConst1 to a precision calculator and the answers should be identical. Always go to full precisions for Area. Amount, summaries etc when testing CFs, even if you are only reporting to 0 decimal places.

    If testing a CF like A.%..AVE(Amount)*CConst1*CConst2 then label several samples in sample set as A plus test the entire formula on at least 3 peaks to confirm its working correct. For validation, depends on in-house SOPs but generally a good practice to get a second trained person to double check the logic of your formula and the results, plus always test in a designated test project not a live project.


    If all ok, copy into Live projects and work away..

  • Hi

    Thanks for your reply

    So to test CF you need to process the sample set

    I am trying to use ENUM data type for Pass Fail type criteria but doesn't seem to work . I don't want to hit 'process' until i am sure it will work


  • Yes, I would always check results or result sets when checking CFs because raw data can be changed whereas a result is a stored and definite figure. Your search order comes into play here. Most custom fields search order is set to Result Set Only, and only custom fields which don't use summaries can show the results in review stage eg a cf of Height/CCalRef1[Height] will display the result when reviewing the sample set as long as you have the processing method open and you have defined CCalRef1. Other custom fields like A.%..AVE(Area) wont calculate unless its batch processed- so no results in review but when a result set is generated the results are there. If you don't normally process sample sets then you could set your order to Result Set First or Outside First (you MUST have a label or a "%" for a label for these search orders to work, though), then process some channels to get results but you wont get the result until the last injection of A.


    So in that scenario, if you processed some injections or channels to generate results, you get a rolling summary with the final result only populating for the last injection of A. Batch Processing ensures the field is calculated and populated for all injections, bar the use of a special field to exclude certain labels or sample types. For example, a CF like EQ(Label,"A")*A.%..AVE(Area) would only populate results at Line A. Enum CFs can be tricky, whats your formula? Do you have the correct sample and peak types included? Is it Peak Enum or Result Enum? We need a bit more information to help.. 

  • Hi ,

    I think I follow

    An example CF would be a mean calculation

    SAME.1..AVE(Assay)

    I cant see any results been generated in the column when I calibrate and quantitate  

    Does this require the sample set to be processed ?

    ie Calculation criteria set at Result set first or only

  • Empower2018
    edited April 2019
    Yes, summary custom fields (those using AVE, MIN, MAX, SUM, PROD or %RSD in the formula) require batch processing to generate results and usually the Search Order is set to Result Set Only. Result Set Only is also the only search order that will generate a result when a CF has no label, for example .%.AVE(Area). All other Search Orders require a label (or a % in its place). If you used your CF with a search order of Result Set First, it would collate all the assay values from the 1st injection of samples with the same label throughout the whole project, which I suspect you don't want? Result Set Only keeps all the results relevant to only the result set. Result Set First, which is the default, first searches for values within the result set before going on to all the other results in the project and including these in the final value. 

    The only way to set up a CF like the above and see the result in the review window of injections or channels is to create a CF like (A.1.(Amount)+B.1.(Amount)+C.1.(Amount))/3, labelling the samples you want averaged as A, B and C, its also advisable to activate the All or Nothing tick box too, that way you only get the desired result at label C, when all arguments are present, rather than nonsensical intermediate values such as A.1.(Amount)/3 or A.1.(Amount)+B.1.(Amount)/2. 


    A simple and basic CF like Area/CCompRef1[Area] or Amount*CConst1+12 can be generated instantly in Review, providing the correct parameters to calculate amount are present. 

    Please note, in order to activate summary custom fields, you must put the Summarize Custom Fields function as the last line in the sample set before you batch process. 
  • Thanks for the information

    So as part of validation of CF what is recommended ?

    Create all CF required (varying from Peak Real , Boolean , ENUM including AVE, % RSD etc) in a 'test' project folder or test them individually?

    Make sure CF search order is Result set only and (fingers cross) batch process ! (I have a large sample set covering many possibilities for CF testing)

    Label standards and samples correctly with calibrate /quantitate , summarise custom fields ?

    On that point does 'summarise CF' require a label reference or leave blank ?


    Thanks

  • It depends on your own requirements. Generally, its better to create and test the custom field in a dedicated Test folder where you can afford to process large amounts of data repeatedly. The search order again goes back to your company and the tests it does. I always use Result Set Only for summary CFs because I want all the results in one set but that might not suit some places.


    When checking the CF results against your calculator or Excel, always go to max precision for real numbers so for example Area*CConst1 would be 102344.45678881092876 X CConst1 to ensure it works. You don't have to report at 14 decimals just validate. For the label column for summarize custom fields, leave blank to include all samples above this line or if only specific labels then include these in the label column.