OR WAIT null SECS
Do you produce data or information?
Do you produce data or information?
As regular readers will know, I sometimes use this column to discuss the “industrialization” of chromatography. In the last instalment I described the difference between a Chromatographer and a Chromatography User (1), and this instalment will continue this theme by considering the following question: What do you do all day?
The response will vary depending upon how cynical you are about your current position, and I guess which industry you are working in.
My question is serious though and could be restated as “What are you paid to do?”
I’d postulate that there are two correct answers to this question for the modern chromatographer: to produce data or to produce information. At first these may seem very similar, but I believe they are very distinct, and the latter may be much more valuable to your employer and enjoyable for you.
Instruments are black boxes in which the sample is placed and numbers hopefully emerge from the other side. Ok, we may need to do a little sample preparation and prepare the instrument, but this will typically involve checking the instrument’s qualification status (in regulated environments), loading a pre-set method into the data system, and perhaps running some system suitability samples and a calibration curve (where appropriate) to check the results against the values and limits within the method protocol. We may load a sequence that allows us to run the samples in the correct order to enable quality checks to be performed postâanalysis. Once the data is collected, we might review the quality control sample data to ensure agreement (variability) within and across the run and that each sample is within specified control limits. We may eyeball the results for each sample to ensure that they “look ok”, which generally means that we check to see if they are within, or out of, specification. For samples out of specification, we may investigate the chromatogram to see if the separation looks “normal”, the baseline is OK, and the integration is sound. If so - that sample will be flagged up and, depending upon your quality management system, submitted for repeat analysis. Then, on with your next job.
So what’s written above is a perfect example of data production. It happens in many companies all over the world in which the “process” has overtaken the science, and you are there as a body, a cog in the machine, which produces data to feed into the big data monster so that decisions on what do next in the project, what to do with that batch of product, or whether to move forward with the next synthetic stage can be made. Your role as a data production robot is therefore invaluable. The company literally couldn’t do without you. Oh wait, actually they could, because there are lots more folks out there just as willing to produce the data to the same quality, using the same protocols and quality management system.
However, for those of you who need a “job” rather than a career, and to be paid commensurately, and take pride in producing data, then that’s just fine. Actually, it’s the way I started my own career. However, if you’re taking the time to read this column then I’m guessing that what’s written above doesn’t fit your personal profile. Or at least you aspire to do more than just produce numbers.
So let’s look at the knowledge skills and attitudes that are typical of someone who is capable of producing information as well as data.
One of the most important distinctions is the amount of knowledge and insight into where the samples have come from, the importance of the information being produced, and how this information will (or can) be used once it’s produced. Perhaps you are even involved in interpreting the data in context and making suggestions or decisions based on your analyses. Of course, this brings greater insight into the requirements for your analysis - perhaps a greater attention to detail or even data accuracy and, without doubt, more job satisfaction. How do I know? Because I’ve seen it so often during my career. Insight and understanding leads to better quality information and better job satisfaction. Period.
Here are some contrasts from the data production laboratory worker that I described above.
I described the typical process for reviewing data post-acquisition and talked about reviewing a chromatogram for a sample found to be out of specification. Why didn’t the chromatographer check all of the sample chromatograms? Perhaps one of the samples had been poorly (incorrectly) integrated and, if it had been integrated properly, the result would have been out of specification. Are we believing in the data without being interested in “information”? The information gained from scrolling through a whole campaign or batch of samples can be invaluable. Changes in baseline appearance, peak shape, retention time drift, and resolution between critical peak pairs can, with the right training and experience, bring a huge amount of insight into “why” the data is being produced and how it fits into the overall workflow. This greater insight will also help you to contextualize your information and also bring you a whole lot more job satisfaction.
Someone who is producing “information” will probably know the analyte, eluent, and column chemistry and will be able to use this information to resolve issues with the analysis. Retention time drift or irreproducibility, problems with resolution related to selectivity changes, peak shape issues, and baseline problems can often be traced back to the “chemistry” of the separation. Often data producers know very little chemistry and sometimes virtually nothing of the chemistry of the separation they are conducting.
Furthermore, chromatographers who are interested in producing information are very aware of the limitations within the “number” they produce for each sample. They will be aware of the type and validity of the calibration approach being used and the way in which the quality of the analysis is being measured, via quality control (QC) samples and cycling recalibration, for example. They will also be aware of the error (or measurement uncertainty) associated with the results being produced. I often ask people working in QC environments why they don’t have a range for their results (50.0 ± 0.12 mg, for example). I often get told that the uncertainty is built into the “specification range” of the assay, which is a valid approach but one that will result in a tighter specification than is perhaps necessary and doesn’t allow the end user to appreciate that any number produced is in fact a range rather a datum. I’m also told that “all that was taken care of during validation” - wow!
I presume that producers of information are more aware of the instrumentation and software, and so troubleshooting a problem or optimizing the performance of the hardware through proper setting of the more esoteric parameters (which may not be mentioned in the method specification) or optimizing the speed of data analysis and report production is possible.
Any data producers who have read to this point may no doubt be thinking: “I don’t have time for all of that, I’m under pressure to get results out of the door”. They should perhaps consider whether the results are meaningful or just data for data’s sake. Valid perhaps within the confines of the methods, specifications, and protocols, but generated with little insight. The fact of the matter is that more juice is being squeezed from each laboratory lemon these days, and if you want a rewarding career in chromatography you need to do two things. First, train yourself by going the extra mile to understand some of the things that I’ve outlined above (sometimes you may need to do this in your own time). Second, you need to insist that your employer supports you with professional development through coaching and mentoring internally, as well as training from external sources.
And just to finish - I call data producers chromatography users and information producers chromatographers. Which are you? Which one do you want to be?
Incognito, The Column12(21), 11–13 (2016).
Contact author: Incognito
Are You a Chromatographer or a User of Chromatography? The Column 12(21), 11–13 (2016).
Thanks for your articles, which I always aim to read, if nothing else in the Journal. I was just reading your latest of 23 November when a PhD student asked me why had his LC retention time shifted? He’s not my student, but I give the students benefit of my experience (whether they like it or not!). I think he was hoping for a straightforward simple answer, but I made a number of suggestions after questioning him further and he’s gone to ponder. Last week another of the students had a “carryover” problem on the LC autosampler. Her Supervisor called in the Service Engineer to sort the problem and I did assist (?) with a few suggestions. The Engineer was good I thought, fresh from the latest training and the student, a bright East European girl, showed interest in solving the issue with him. I think this is part of the problem now, that instruments are under service contract and it’s too easy just to call in the Engineer to fix things. In the past, solving the problem would have been a valuable exercise for the student and part of the “fun” of research.
There are other examples I could bore you with, but one other I have to mention is when I took over running a laboratory for a Professor of Medicine and Nutrition at the current number 1 university in the “League Tables” (don’t get me started). The laboratory LC–MS instrument had been run by one worker for the previous 4 years and she took it very unkindly when I asked her about the mass calibration on the MS. She had never calibrated it, leaving it to the Engineer, at least once a year! She had been producing quantitative MRM data for several papers over these years and I thought she was generally a good worker, but clearly lacked sufficient knowledge. Maybe precise but not accurate data.
In answer to your request for suggestions, after this long preamble; I don’t really have any and think the situation is not very hopeful at present. I can foresee some serious mishap, possibly arising from drug development and lack of suitable chromatographic or other knowledge, before things might change. Sorry to be so pessimistic.
Please keep up the good work. - A Reader, UK
Thanks for that nice article about chromatography users and chromatographs.
Well, I’m not really sure if there’s a certain border between “chromatography users” and “chromatographers” and if you can clearly separate them into two groups.
There’s also a grey zone between computer users and computer programmers. You might be an expert in MS Office including Visual basic scripting, writing perfect VBA add ons and never heard of C/C++/C#/FORTRAN. You might also be fluent in a couple of programming languages on various operating systems, but never produced anything fancier than ASCII text files.
I dare to call myself a “chromatographer” with more than 25 years of experience (master thesis “graphic-assisted factor analysis for peak deconvolution via UV diode array spectra”, 14 years in instrument development for the petrochemical industry, working with LC, IC, SEC, GC), but that’s nothing very special.
I completely agree with you, that most important is the spirit, the challenge of separation problems, squeezing the most possible out of existing hardware and, of course, finding a sensitive, rugged, and stable solution for an upcoming problem. A solution which can be used and understood(!) by people maybe not that familiar with the unit. And I love my job.
And the big number of pure “users”, well, I do believe that we need them. Chromatography has occupied that much of daily lab work that the “chromatographers”, as you call them, simply can’t pick that burden alone.
In our lab I was (lucky me) able to inspire my colleagues for new methods, new practices, and various optimizations. Of course, many lab employees will refuse to shorten or replace a capillary column by themselves, pimping up an FID, creating special methods, or running some experiments on different eluents. They won’t recalculate Van Deemter curves, theoretical plates, or diffusion terms in packed columns.
But even my 11-year-old son understands what chromatography is and how it works.
(My beloved example for younger people: “Who is interested in clothes? Cinema? Food? All together? Nothing? Imagine a giant mall with shops, restaurants, and cinemas. Ten thousand of you walk through the mall and may stop where they want. Who will arrive first and who latest?”)
I’m sure you cannot implant some enthusiasm for chromatography, but my opinion is (for nearly every part of natural science) that you can guide students and graduates to a higher level of interest for separation science (it’s necessary everywhere), as long as that (definitively quite complicated) stuff is carefully dosed. Math is taught in nearly every course, but, sadly, analytical chemistry only in a few.
Just my two cents. - A Reader, Austria
Time for a Change? The Column 12(19), 8–10 (2016).
I agree that it is well past time for a change in my lab, but I am constrained by factors I cannot control: business and cost of equipment. Our company derives all of its income from catalog sales, and although business is growing, investment is not being made in the analytical side but in the R&D and manufacturing sides. It’s darned depressing some days.
Thanks for your column. I appreciated the thought in it and the sincerity that motivated it. - A Reader, USA
The discussion regarding the challenge of change was interesting, but missed one of the key inhibitors of change within much of the analytical community. Assays which are part of fulfilling regulatory requirements are both difficult and costly to change. Once a method is “validated” and meets regulatory muster, the cost of change in time, money, and effort is a very large hurdle that is difficult to overcome. The most recent guidance document from USP on chromatographic methods USP <621> rather emphatically makes it clear that changes in gradient methods are not encouraged ...”Adjustments to the composition of the mobile phase in gradient elution may cause changes in selectivity and are not recommended. If adjustments are necessary, change in column packing (maintaining the same chemistry), the duration of an initial isocratic hold (when prescribed), and/or dwell volume adjustments are allowed.”
The theory and practice of modifying gradient methods to preserve L/particle diameter and to keep the number of column volumes constant for each gradient step are well understood, but in practice the majority of QA or QC laboratories are unwilling to change gradient methods to take advantage of the higher resolution and peak capacity available with columns packed with sub-2-µm particles. The perceived cost of revalidation is a major reason why many laboratories persist in using outdated methods which do not provide data in a timely fashion to meet the real business requirements for efficient manufacturing. I recently encountered a situation in which a formulary gradient method required 2 h of run time per injection and a demonstration of an acceptable RSD for six injections of the standard before samples could be run. This virtually ensured that analysis of a batch of an API would not be available within less than 14 to 16 h following collection of samples. The cost of holding large batches on an API in quarantine was felt but not amortized. All of the System Suitability metrics could easily be met with a gradient separation requiring less than 20 min of cycle time.
Clearly changes in antiquated regulatory expectations will be needed as a more business goals centred view of analysis is developed. The correct questions are ... does this assay provide the information necessary to assess quality and take corrective action (if needed) on a timely basis? Is that information more valuable than the cost of revalidation of an antiquated method? - A Reader, USA
For many years, the Incognito column has been educating, challenging the status quo, and gazing into the future of the field of chromatography. Whilst there is never any shortage of topics to write about, it’s always nice when we get feedback from our readers, which shows that you care enough to take some precious time to give us your opinions or advice on the subjects we tackle.
I clearly remember back in 2009 when we wrote about the great acetonitrile shortage and what one could do to overcome the problems this created. The number of responses almost broke our e-mail inbox with requests for a calculator we had built to come up with elutropically equivalent solvent mixtures. Whilst this has never been repeated, I’m always happy to see reader feedback and will try to respond whenever possible.
Readers’ comments are often very insightful and add a lot to the arguments that are being presented. I’ve therefore selected some particularly noteworthy examples to share and expand upon the information presented in the original column.