Wendy Warr Interviews Alex Tropsha

WAW: A recent letter to the editor of J. Chem. Inf. Model. reports, “A general feeling of disillusionment with QSAR has settled across the modeling community in recent years” (Johnson, S. R. The Trouble with QSAR (or How I Learned to Stop Worrying and Embrace Fallacy). J. Chem. Inf. Model. 2008, 48, 25-26). How do you feel about that?

AT: To be honest, I am not quite sure which community Steve Johnson was referring to, since my many colleagues and I who (unlike Steve) have been working seriously in this field for many years are most certainly not part of such a community. I was quite surprised to see this letter accepted for publication in its final form. I expressed my views in an e-mail to about a dozen of my close colleagues who were quite flabbergasted to see it published. In brief, in my opinion the content of the letter fails several simple tests on scientific rigor and in fact is rather pointless. For instance, pharmaceutical companies synthesize tens of thousands of compounds every year in a hope that each of them will be at least biologically active, if not immediately becoming a drug, but experimental hit rates are generally low and successes are rare. Nevertheless nobody (especially a non-specialist who does not work in the wet labs and consequently should not have the right to criticize experimental chemists) takes upon medicinal chemistry as a whole just because one medicinal chemist got “disillusioned”.

I am always highly appreciative of any constructive criticism of my field (and several important papers published in recent years have seriously examined various problems in QSAR field, e.g., outliers and activity cliffs discussed in a letter that Gerry Maggiora published in J. Chem. Inf. Model. last year). Steve Johnson’s letter states several problems that have been known for years, and that the QSAR modeling community has been addressing for at least five years, but it offers no new insights. So my feeling (that I know I share with quite a few of my colleagues) is that of disappointment that such a low quality paper was published. On the positive side, it may stimulate really serious discussion about the quality of the field and where it is going.

WAW: The International QSAR Society became the QSAR and Modeling Society and is now the International Cheminformatics and QSAR Society. You are now vice chair of the society. Why the change of name?

AT: I commented above on my view of QSAR modeling as a field and a tool but not as a broad research discipline. So, Tudor Oprea, the society’s chair, and I thought that in modern days QSAR is part of a broader discipline of cheminformatics. We had had many discussions among the members of the board of the society on the issue of name change. We finally agreed that with many members of the society actively embracing cheminformatics, and with growing activities in the area of cheminformatics research (e.g., formal courses and even degrees given in several universities around the world, and funding for cheminformatics established by the NIH RoadMap program), the inclusion of the word “cheminformatics” in the name of our society is more than justified. At the same time we felt that in part for historic reasons, in part for the reason that QSAR remains a major quantitative methodology for biomolecular data modeling, it should stay as part of the society name as well.