Is there any way to calculate average amount for samples run with different calibration curves?

For samples that need to have a bracketing standard every 5 samples and is quantified using sliding bracket with overlap, I cant get an average amount custom field to work out. So if I run a standard, 5 samples and a second standard, then 5 more samples, another standard etc etc I cant get an average custom field (SAME.%..AVE(Amount) to work if the samples of interest are split over 2 calibration curves, and the method is validated for this bracket set up so I cant change it. 

I know I could use the report method to calculate specific averages but I need to perform a further Bool function on the average so I need an actual value. Is there any other way I can calculate this and still have the amounts calculated from different calibration curves?

Answers

  • Assuming you already have a data set for which you'd like to do this...
    Go carefully through the sample set wizard and choose that standard calibration option. Look at the resulting labels, processing directions and cal/quant lines and add/insert/modify your extant sample set to match, then reprocess.

    You do not need CFs for any of this. You will essentially be using the same calibration curve at several different points in its creation, each of which can be tagged in the report method with the addition of a cal ID field.
  • I hear what you are saying Dan but my issue is how do I bypass the fact the method states to quantify the 5 samples off the bracketing standards. When I try something like, for example, A.%..AVE(Amount) and split the A samples such that 3 in one group of 5 and 3 in the second group of 5, it doesn't work right as the lines "Calibrate" and "Quantitate" seems to stop the flow of averaging. 
    Im not in front of Empower for a few days but does anyone know if this only happens when averaging peak values such as Amount? If I had a custom field that calculated Amount*CConst1 and I wanted the average of THAT custom field over several calibration curves would it work? My logic is the Amount*CConst1 is already calculated before the summary kicks in..
  • shaunwat
    edited February 2019

    I am having a hard time getting a clear sense of what you are after. Sounds like system suit, bracketing standards, and calculate samples that run between those standards. So....

    STD 1

    SMP1

    SMP2

    SMP3

    SMP4

    SMP5

    STD 2

    Where Samples 1-5 need to be calculated by Standards 1-2.

    That's easy to do if you only have one set of samples. The bear lies when you have to calculate Samples 6-10 from Standards 2-3.

    Though I am not fond of their use, you might be able to get something to work with the CConst1 approach. However, I am a little lost as to what you might be able to throw into CConst1 to make that calculation work.

    You may be able to create a custom field for each of your standards and manually walk the sample calculations through an amount result via labels and custom fields. You would need to anticipate maximum brackets and code this number of result custom fields.

  • Hi shaunwat and thanks for the reply. Ill try to explain my question more clearly. Yes, the bracket set up you mentioned is correct. Lets say I'm running Batch A and Batch B- Batch A comprises of 10 individual parts and Batch B also comprises 10 individual parts. My validated method states to run 6 samples then a standard, and to use the bracketed standards on each side of the samples to quantify these samples.


    I need to get an amount from these samples. My issues is I'm running the batches as follows in the sample set:


    Standard 1

    Batch A Sample Part 1

    Batch A Sample Part 2

    Batch A Sample Part 3

    Batch A Sample Part 4

    Batch A Sample Part 5

    Batch A Sample Part 6

    Standard 2

    Clear Calibration, Calibrate, Quantitate

    Batch A Sample Part 7

    Batch A Sample Part 8

    Batch A Sample Part 9

    Batch A Sample Part 10

    Batch B Sample Part 1

    Batch B Sample Part 2

    Standard 3

    Clear Calibration, Calibrate, Quantitate

    Batch B Sample Part 3

    etc etc

    I need to get the average amount for each batch. So I code a custom field called SAME.%.SAME.AVE(Amount), and label all parts of the batch the same. But this wont calculate correctly. All the parts of Batch A are split over 2 calibration curves and it doesn't give the right answer. When I tried to bracket all 10 parts of Batch A between 2 standards, it works perfectly. But I'm tied to this 6 samples per bracket layout so I was wondering is there any way around this, ie to still get the average amount and maintain the sample set layout?

    However, after trying this out a few times yesterday I have accepted that it just wont work that way, unless I'm missing something. Either all samples required for the average need to be in the same bracket or you need to change your standard layout to injecting every 6 samples but only quantitating at the end of the sample set.

  • What if you use "Don't process" for all injections and Clear, Calibrate, Quantitate at the end?

    STD1
    BATCH A 1-6
    STD2
    BATCH A 7-10
    BATCH B 1-2
    STD3
    BATCH B 3-8
    STD4
    BATCH B 9-10
    STD5
    Clear Calibration
    Calibrate STD1/2/3
    Quantitate BATCH A*
    Clear Calibration
    Calibrate STD2/3/4
    Quantitate BATCH B*

    Just curious, is this a Ph. Eur/USP method mess?
  • Hi DavidHPLC, yes that set up does work when I tried to get an average amount for each batch (10 stages per batch) but my problem is my validated method states to quantitate the 6 samples "from the average of each bracketed standard around it" which means the 2 standards injected before and after the 6 samples. It would be better to just set up one big calibration curve with all the points at the end but I cant change the method.!

    Its an interesting one though, I really didn't think Empower would even consider the set up of calibration curves for calculating average amount. I presumed the same label would be enough to go by but apparently not. 
  • Are you utilizing label references in your calibrate and quantitate functions?
  • Hi shaunwat,Yes I use them as follows:

    Standard 1 (S01)
    Sample #1 U01
    Sample #2 U02
    Sample #3 U03
    Sample #4 U04
    Sample #5 U05
    Sample #6 U06
    Standard 2 (S02)
    Clear Calibration
    Calibrate (S01 S02)
    Quantitate (U0*)
  • shaunwat
    edited February 2019

    Hmmmm.... I am not sure if Empower is going to understand the Calibrate (S01 S02) label reference. I believe Empower will literally look for a label "S01 S02".

    Though I understand this will cause problems for you down the line try the following:

    1. Only have labels for your lines shown above and leave the labels as you have above.

    2. See if your calibrate and quantitate functions work properly. I suspect no.

    3. Change calibrate to: Calibrate (S0*) and see if that works. I suspect yes.

    The problem again becomes trying to get that wildcard calibrate function to work in the manner that you need. It's really the fact that your bracketing standard injection is used both to close the previous bracket and open the upcoming bracket.

    Because of the dual function of the bracketing standard it is going to be nearly impossible to come up with a label convention that can be both used in the previous and upcoming bracket, which can also be triggered by a calibrate function.

    EDIT:

    I'm also not sure if the full * wildcard can be used as a label reference. You also may want to try a variation on the steps above, but use the single character wild card (?). For example, calibrate (S0?) and quantitate (U0?).

    I seem to remember having problems with the full * wildcard when used as a label reference.

  • DavidHPLC
    edited February 2019
    I have to say I never had problems using multiple labels and/or full wildcards for Calibrate/Quantitate (using Don't process for injections and Normal for label instructions).

    Empower2018, can you post the screenshot from Empower of what you tried to do in the February 12 post (obfuscate sample names if possible)? With the combo of labels, label references, type of processing it's hard to get the whole picture.


  • @ Shaunwat, I never had a problem using either ? or * in my label columns, for example I run Std1 and S0101 and Std2 and S0201so my calibrate label is S01* S02* and it recognises the right labels and calibrates accordingly. Same with unknowns. 

    @DavidHPLC, ill try to post this tomorrow. 

    Im not sure my point is coming across. Its not that I have a problem calculating amounts for all these samples using this bracket set up, the problem is a custom field of the type SAME.%.SAME.AVE(Amount) doesn't work properly, it wont give the right answer UNLESS all the samples you want averaged are all together between the two standards. If they split up over 2 curves, then it wont work. I wonder is it to do with the fact that sample sets are processed line by line and when Empower sees the calibrate function several times whilst trying to calculate the average, it just cuts out.