How do you do method development?

[Alfred88]

Dear Tom (or someone with LC Resources):

Could you tell about the current pricing of the Drylab software (one license for Windows OS)? Any demo software available? Any way to get demo copy of Drylab, ChromSword or others?
Thanks.

Alfred.

[Yury Zelechonok]

I am working for a column manufacturer. Since our company’s columns (Primesep and new line Obelisc) are non-traditional RP, IE, NP columns, neither book of Snyder nor Drylab can provide any practical help. Never the less we do many methods for clients (free of charge). We usually start when all traditional MD approaches are failed.

Uwe-
do you want to know how we do it or you thread has the purpose to see what is going on within “traditionalâ€

[Uwe Neue]

Hi Yuri,

I thought that the discussion of successful method development approaches would be useful for everybody, and since RP is the most popular technique, a discussion of RP techniques would be the most promising.

I personally think that combining RP with IEX makes things more complicated. However, I have not yet had a direct experience with method development on your packings, and I am sure that we can all learn something from you. So: feel free to contribute!

[Mattias]

A comment on the Dry Lab software: I don't think it matters for the software what kind of retention that you have. From what I have heard I just draws linear models from the test experiments. That means that you should have a fairly good idea about the pH and ionstrength of your method before you start using Drylab for optimisation. (e.g. you cannot make two experiments at pH 2 and 7 and then find an optimum pH of 4.6)
Anyway, I think Primesep columns and Drylab should work just fine (although I have not tested it yet)

With regards to the original post, since I work with peptides the parameter I use for development is mainly pH. For my work the difference between different columns (brands, C18, Phenyl, CN etc) is very small compared to the effect that pH has on my separation. So I usually take make favourite RP column and do the pH optimisation on that. Right now I favour the Zorbax Eclipse C18 that I found to be very reliable, with the best plate count and peak shape for my applications. But that varies over time. Lately I have started to use the sub 2 µm columns with great success (on my standard HPLC).

[Uwe Neue]

I have not worked with DryLab, but knowing the creators, I have strong doubts that it just interpolates along straight lines. RP is a log/linear relationship, and the relationship of retention with temperature is a 1/T relationship. I do not know what Drylab does with pH, maybe Tom can comment on this.

One of the things that we did with peptides is to change the pH, going from acidic to basic. You get MUCH change in elution order and retention. Plus, polar peptides you can do with HILIC.

[Rob Burgess]

To Alfred88 and anyone else who may be interested,

Correct me if I am wrong on this, but isn't DryLab software being handled by the Molnar instiutute now....

[HLR_gutc]

For more information about Drylab please visit the Molnar-Institut homepage.

[Uwe Neue]

There were a few posts where I had hoped to get additional info. For example, how to do method development on mixed-mode phases, or how people use the software packages that are out there.

Do they work? There are many publications by Dolan and Snyder that show that Drylab works, but how are these tools in the hands of less sophisticated users? Other softwares are available as well (e.g. Galushko)? How do they compare?

Or are we so good that we do not think that we can benefit from method development software?

Any comments?

[juddc]

Interesting topic - sorry I'm late to the party!

I run an analytical lab servicing an R&D group making prototype raw ingredients and finished formulations for cosmetics. What I do is quickly learn what I can about the analyte (solubility, pka, chromophore, fluorophore, etc...), then get as much information on the matrix as possible in order to guide sample preparation. If possible, I like to obtain an accelerated stability sample that's been suitably abused to run along with standards initially.

Obviously, I tailor detection to the properties of the molecule (or derivative, if necessary), but I'll also frequently run an ELSD in line initially just to be sure that I'm not leaving hige quantities of lipidic goop on the column...as long as I don't need a phosphate based MP.

Column selection evolves as the method does. My default is usually a base deactivated silica C8. If selectivity is an issue, I'll first vary the organic (MeOH, ACN, THF usually) then pH if I can get away with it. Stability at higher or lower pH and necessity for detection at low wavelengths dictates this.

Then I'll work with different supports depending on the nature of the analyte and suspected interference. If more retention is needed, a C18 on similar silica is used. Sometimes a phenyl or phenylhexyl column, sometimes one with an imbedded polar group, again depending on the analyte / matrix.

If none of this works, I will not hesitate to use an ion-pair reagent or even ditch reversed phase completely and go with an amino or silica column if there's an argument for it. I've had luck using an IP agent for an analyte that didn't require one, but was in a matrix that did: Ascorbic acid in cell culture media

I do use automated method development software (Waters AMDS) where relevant, usually initially to get some preliminary separations done unattended so that I can use instrument time more effectively and to help guide decision making.

Later on, I'll use the modelling portion of the software to optimize band spacing and column geometry so that I can have a more efficient run to hand off to our QC group. I've found it quite handy to vary column length, diameter, particle size, and gradient rate electronically, then buy the column in dimensions that make for the most efficient separation.

Hope this helps...

[tom jupille]

I have not worked with DryLab, but knowing the creators, I have strong doubts that it just interpolates along straight lines. RP is a log/linear relationship, and the relationship of retention with temperature is a 1/T relationship. I do not know what Drylab does with pH, maybe Tom can comment on this.

Mea culpa for not following this thread closely enough (I really wanted to listen more than talk, and then I've been traveling most of the past month).

DryLab is now officially part of Molnar Institut.

The fitting functions vary with the type of variable:
- isocratic reversed phase is log(k') = log(k0) - S Φ
- optional fit to quadratic of log of Φ (for systems with lots of non-linearity)
- gradient reversed phase is "linear gradient model"
- temperature should be log(k') is function of 1/T (in deg K), but log(k') as linear function of T (in deg C) is close enough.
- pH was (in earlier versions), log(k;) as a linear function of the % dissociation; in current versions, it's simply a cubic spline (less elegant, but more robust!).
- ionic strength, buffer, additives, etc. is log(k') as a linear function of log (ionic strength).
- All the isocratic modes will allow additional data points (and usually revert to some form of cubic spline fit)
- the "two-variable" modes are linear combinations of the above (temperature/gradient time is the most widely used).
- there is a neat "custom" capability which lets users define their own models with linear, log, quadratic, cubic spline fits in various combinations.

I can't speak directly to the other modeling programs (ChromSword or ACS Chromatography Simulator), but I think they use pretty much the same algorithms; Lloyd Snyder has always been open about publishing.

The biggest obstacle with any of these programs has always been "peak matching": you have to match up the corresponding peaks in the various calibration runs. At PittCon, Molar showed me his "Peak Match" program which helps a lot with that. The user interface is daunting (4 chromatograms and an array of buttons), but it's a lot more effective than "seat of the pants".

As to usability by "less-experienced" chromatographers: absent the peak-matching issue, it's actually pretty good. If nothing else, it forces a systematic approach to method development; the user has to decide what variables to address and then explore their effects rather than rely on intuition or whatever worked last week. When we bought the LC Resources training business back from Rheodyne at the beginning of 2004, we specifically retained the rights to use the DryLab software in our training courses; there really is no substitute for having to immediately apply the contents of a lecture!

[mk12]

I have to give interview presentation in a pharmaceuticla company. I wanted to show how i developed method and validated it. Can I use method developed for one the drug in my first company to another company without telling the name of the drug. Can I use the data without telling the name of the product . Please help asap cause my interview is on monday

[syx]

Dr. Neue, I hope I am not late to answer your initial post in this topic. I work in a pharmaceutical industry.
We always search pharmacopoeial method and then published method first. If we get them, we compare the methods each other and try from the simplest one. We perform specificity test with our placebo. If it is ok, we continue the test with other steps of verification or validation.
If we could not find any literatures, we collect all information about the substance: molecular structure, acid/base/neutral, pKa value, chelating properties, polarity.
Next, based on those data we choose pH and column that may be suitable for the substance. The initial organic solvent is acetonitrile. The method is optimized using stess test and specificity test.

[tom jupille]

Alfred88:

We still use the program in our training courses, and I occasionally teach a course on how to use it, but LC Resources has no commercial ties to the DryLab software. As of early 2007, DryLab "belongs" to Imre Molnar (http://www.molnar-institut.com). There is an evaluation version available, but you have to e-mail a request to get the link. I would assume he's raised the price, but I don't know what the current pricing is.

I think the US dealer is still Dave Isom at Analytical Sales. His e-mail is david@analytical-sales.com, or you could try phoning him at (800) 899-4752.