1. bronxite on October 19, 2005 9:33 PM writes...

Have a look at "The Value of R for Pre-Clinical Statisticians" in this newsletter. I get a little queasy when I see math and discovery (try to) mix at biotech and pharma. Computational chemistry has certainly by now been "institutionalized" within this realm. Even DE Shaw invests in it.

Some people have actually tried to tame these theories by integrating observation, but the endemic "physics envy" in the field makes this a slog.

I shouldn't be too hard on these people. Some bright young sparks seem to be recognizing the value of phenomenology. I just wish that statistics was as institutionalized within this realm as computational chemistry.

2. LNT on October 19, 2005 9:37 PM writes...

If only the ACS would have a publication that accepted medicinal chemistry papers without experimental details... They are really loosing out on some good science because of J.Med.Chem's silly requirement for elemental analysis. Personally, I think that J.Med.Chem has to accept all these "virtual" medchem papers because they make it so darn hard for "real" chemistry to be published in it!

3. Petros on October 20, 2005 2:51 AM writes...

It has always been thus with J Med Chem and, prior to the emergence of BMCL, there was little option of publishing Med Chem papers elsewhere.

In the major company I used to work for microanalytical data were never obtained thereby precluding publication of chemistry papers in a major journal. I remmeber my surpise when some colleagues published a short paprer in J Med Chem, describing compounds prepared in mg quantities.

They had shortciruited the journal's requirements by supplyig fabricated CHN data!

4. bronxite on October 20, 2005 8:30 AM writes...

Why the resistance to elemental analysis? It's simple, no? Is it just a foregone conclusion in any "professional" experiment? If so, why are JMC such sticklers?

5. Derek Lowe on October 20, 2005 8:59 AM writes...

Elemental (combustion) analyses are a pain on several levels. For one thing, they destroy a lot of material, compared to lossless techniques like NMR or the microscopic amounts needed for LC/mass spec. Another problem is that they're rarely done in-house, so you have to go to the trouble of sending them out.

And the biggest problem, as far as I'm concerned, is that they're just not worth much. They're useful if you're wondering if you have a hydrated compound (although a good NMR will tell you that), or if there is some metal contamination. But for routine purity checks, we all have too many stories of samples that failed combustion analysis, were mailed right back in the same vials, and passed. I've done it myself.

6. Drew on October 20, 2005 11:14 AM writes...

If the CHN is spot on, you have a watertight purity; there simply cannot be any contaiminants (salts, silica gel, etc). Many of these are invisible by NMR or HRMS, so ACS feels that the proof in the pudding in CHN anaysis. However, you can have very, very pure material and still fail CHN for reasons unknown....

7. Milo on October 20, 2005 11:40 AM writes...

I do think that a little bit of modelling in a med chem paper is a good thing, provided you don't rely to heavily on it. Chemistry is a 3-dimensional science, and I have a hard time seeing how a carbonyl H-bonds to a tyrosine in an active site (for example). Especially when that kind of stuff is drawn with Chemdraw (2D). A nice ball-and-stick model, while not real, does make it easy to visualize potential interactions. If it is backed up an SAR, then all the better.

8. LNT on October 20, 2005 12:34 PM writes...

Drew, while I agree with you in principle, I think that the ACS has a double standard. They have a clear forum for organic chemistry papers that lack "bulletproof" characterization (Org.Lett, J.Comb.Chem.) but they lack such a forum for medicinal chemistry papers. CHN analysis is a great thing in principle -- but it's just not practical for fast-paced industrial research.

9. bronxite on October 20, 2005 9:52 PM writes...

Thanks for the elaborations. On the face of it, it seems like the reasoning behind the CHN requirement needs more study. I'm assuming the stories here are based on years of experience. CHN appears to have a finite fault rate that the JMC editors choose to ignore; a double standard if I've ever heard one. And it sounds like its purpose could be well served by an alternate battery of tests. The choice of test could just be stated explicity somewhere in the manuscript.

It's ironic that the editors are so fussy on this, yet have such intestinal fortitute for highly approximate predictions from a calculation with poorly explored heurisitics. The results of incomplete experiments are probably much more valuable in their place.

10. PsychicChemist on October 21, 2005 1:52 PM writes...

Are you guys absolutely sure that J. Med. Chem still fusses about CHN analysis.

We got a letter accepted in J. Med. Chem recently and the only characterization data we had was proton NMR (no 13C NMR) and LCMS data.

Although, I am not too keen on writing a full paper for J. Med. Chem as that would mean putting together an experimental with NMR and LCMS data for 50-80 analogs.

11. Derek Lowe on October 21, 2005 3:44 PM writes...

Here's the relevant section for the J. Med. Chem. guide for authors. Letters are less stringent than full papers, and the two-HPLC system is one way out of the whole difficulty:

"For Articles and Brief Articles, authors should discuss the purity of the target compounds and its relationship to the pharmacological data in the Results section of the manuscript. Adequate evidence to establish the degree of purity must be provided for all new target compounds. Combustion analyses are normally required for new, non-oligomeric target compounds that are important for the interpretation of the relationship between structure and biological activity (SAR). Analyses that agree with the calculated data within (0.4% should be printed in the Experimental Section as follows: Anal. (C23H31N5O2) C, H, N. In cases where it is necessary to rely on combustion analysis data that exceed (0.4% of calculated values, the data will be presented in full in the Experimental Section of the paper, e.g., Anal. (C16H20N2O8S) C, H, S; N: calcd, 6.99; found, 7.55. All combustion analysis data should be listed in tabular form in Supporting Information (see section r). When combustion analyses for key target compounds are not available, the results of two diverse HPLC systems should be reported as criteria of purity. For target compounds that are not critical for the interpretation of the SAR, the minimum acceptable purity is 95%. The purity of such compounds may be determined by combustion analyses, HPLC, or spectral methods. For Letters, the characterization and purity of target compounds should include NMR, HRMS, and HPLC as Supporting Information (see section r)."

12. LNT on October 21, 2005 3:46 PM writes...

It mainly depends on the reviewer from what I've heard. If the reviewer is a stickler for the rules, then you'll have to include CHN -- but you are correct that much of the time I think you can get away without it.

13. heeler on October 22, 2005 10:11 AM writes...

The other oddity of JMC's criteria is that their compound characterization demands (as listed) are more stringent than JACS or JOC. I'm all for quality data, but I'm also aware of publications from a major pharma company that included xray data representations (no coordinates ever depostited) for molecules we and other companies discovered were non-specific inhibitors. It was not very hard to determine this after we went to the trouble of synthesizing the things. I would prefer better looks at this type of data.