Short CV: I got my Ph.D. from University of Szeged, Hungary in 1959 then became a member of the Peptide Research Group of the Hungarian Academy of Sciences. As arecipientof a post doctoral fellowship of the National Research Council ofCanada in1964, I worked for one year in the Departmant of Biochemistry, Universityof Alberta in Edmonton. I joined Eötvös Loránd University,Budapest, Hungary, in 1967 where I held the position of Professor of OrganicChemistry since 1972. Retired in 2002. From January 1995 to May 1999I spent an extended sabbatical at Advanced ChemTech, Inc., Louisville,KY,USA.

  The roots of combinatorial chemistry in our laboratory  go back to 1966 when I returned to Budapest from the post doctoral year spent at the University of Alberta, Canada. Under the guindance of Prof. L. B. Smillie, we determined the amino acid sequence of chymotrypsinogen-B and I was wondering from how many sequence possibilities did we choose the right one. From the numberof aminoacid residues (245) and the number of amino acid building blocks(20) the numberof possible sequence combinations (20²⁴⁵or 5.65x10 ³¹⁸) could easily be deduced. It soon turned out that the estimated quantity ofmatter in the whole visible Universe would not be enough to build up even a singlemolecule of each sequence. According to estimates (See Scientific American1994, November, p. 48) the tota lnumber of elementary particles is“only” 10⁸⁸ . This was my first, and shocking, encounter with the immense kingdom of  molecular diversity.

  After the adventures in this kingdom, it was a logical continuation around 1980, to think about the possibility of  synthesizing all peptide sequences. It was quite clear, however, that using the conventional techniques in preparation of libraries longer than tripeptides would be impossible. My first idea was to use an equimolar mixture of the 20 different N-protected amino acids in the couplings. This would lead - at least in principle -to formation of arapidly growing number of sequences and finally a full peptide library could be cleaved from the support in the form of a mixture. It was clear, however, that in suchcouplings the products would form in unequal molar quantities as a consequence of the differences in the reactivities of the amino acids.The differences in molarities would be amplified in each successive coupling step leading toa mixture with uncertain composition. Rethinking the possibilities ledto anew idea in the early spring of 1982: the portioning- mixing (split-mix)procedure. This completely eliminated the problem connected with the differences inthereactivity of amino acids and the advantages of the solid phase method(thepossibility adding ther eagents in large excess, or repeating the couplingoperation)could be fully exploited.
    A big problem still remained. Ever since the beginning ofmodernsynthetic organic chemistry, the goal of chemists was to prepare singlecompounds in as pure a form as possible. In the modern organic chemistry,producingmulti-component mixtures and using them in the drug discovery process,seemedunacceptable. Forthis reason there was an urgent need to present inaddition,an efficient strategy for identification of the bioactive substancethat maybe presentin the complex synthetic mixture. Fortunately, I coulddevelop atheoreticalsolution in a veryshort time. I called it “syntheticbacksearchingstrategy”which much later proved to be in principle identicalwith the"iterationstrategy",published later by others.
    I was fully aware of the importance of the combinatorial approach in pharmaceutical research but all those I contacted for cooperation showed no interest at al. For several reasons at that time it seemed impossibleto patent it. One of the patent attorneys I was in contact with,  Dr.Éva Somfai, suggested me, however, to describe the method in a document and - inorder to give me some support in potential future priority disputes - notarise it. I did so and the original document written inHungarian- in which the principlesof combinatorial chemistry, including both synthesis and the iteration screening strategy,were first clearly explained - was notarized in May, 1982 .

   In the new synthetic procedure, later called portioning-mixing or split-mix synthesis, the coupling cycle of the Merrifield solid phase method is replaced by three simple operations.

                 1. Dividing of the solid support into equal portions
                2. Coupling a different amino acid (or otherkind of  monomer) to each portion
                 3. Mixing the portions

Then, by repeating this cycle, another amino acid residue is attached to each peptide. The procedure is demonstrated in the scheme of the split-mix synthesis.

In order to facilitate the realization of the synthesis a simple manual device was developed that was successfully used in the preparation of peptide libraries. Later an automatic synthesizer (ACT 357) was developed at Advanced ChemTech Inc.

   The components ofthe synthesizedpeptide mixtures were identified by two dimensional high voltage paper electrophoresis.In order to facilitate the identification, a software was developed. Usingthis    software in a very small computer (Spectrum, 128K) the sequences of theformed    peptides could be generated. Based on the sequences the computer calculatedthe molecular weights, theelectric charges in two different (pH 2, and pH6.5) buffers and the relativeelectrophoretic mobilities of the peptides.Inthe identification process theexperimental peptide maps were compared    tothe    predictedmaps generated by the computer.The software made possible    t   ogenerate    allcomponents of huge peptide libraries.Thesewere the first   examples    what   arecalledtodayvirtual libraries.The small computer worked    for several   days,   forexample,to prepare the predicted electrophoretic map   of the64 million    hexapeptides.

Of course the synthesized mixtures were much simpler than the large virtual libraries to allow identification of all components. The prepared peptide libraries contained 9 to 180 components. The identification is exemplified by a mixture containing 12 tripeptides.

    The split-mix method first appeared in print in the Ph.D. thesis of our Ethiopian graduate student Mamo Asgedom in 1987. He was in large part responsible for experimental realization of the method.

     In 1988 the method was published on two international congressesas posters under the titles: Cornucopia of peptides by synthesis (14th InternationalCongress of Biochemistry, Prague) and More peptides by less labour (10th International Symposiumof Medicinal Chemistry, Budapest).

   A paper was also published in International Journal ofPeptide and ProteinResearch. The manuscript was submitted in February 1990 and appeared in June 1991.

The list of other papers published by our group later in the field of combinatorial chemistry can be found by clicking on the list.

   The split-mix synthesis - beside the many advantages - has two disadvantages when compared to parallel synthesis. Although individual compounds are formed on each bead, their quantity is low and their identity is unknown.The string synthesis is a  modified version of the split-mix method developed at Advanced ChemTech Inc. and was designed to eliminate these disadvantages. The conventional solid support resinwas replaced by unlabeled macroscopic support units in order to produce individualcompounds in multi milligram quantities. It was shown that omission of the tagsis possible if

    (i) the unlabeled support units arearranged into spatially orderedgroups and the relative spatial arrangement of the units is maintained duringthe chemical reactions
    (ii) the support unitsare redistributed between the reaction steps according to a predetermined pattern permitted by the combinatorial redistributionrule and
    (iii) the sorting processis simulated by a computer that can trace the synthetic history of each support unit.


   The simplest way formakingspatially ordered groups is stringing the units. In a practical realization,Mimotopes SynCrowns were stringed on fish line then each string was submittedto coupling with a different amino acid. After coupling, the crowns wereredistributed among the strings of the next coupling step using a simplemanualdevice.

  In order to ensure formation of all combinations of building blocks in the products, redistribution has to follow the combinatorial redistribution rule: each product formed on on any string has to be evenly distributed among the strings of the next reaction step. There are different distribution types obeying this rule.

    When the manual device is used, the semiparallel redistribution type is the fastest choice. This is illustrated with the first redistribution in a trimer library synthesis from 3 building blocks in each step.

   The software thatiswritten in Visual Basic is an essential part of the method. The versionthatcan be applied for semiparallel distribution is accessible in the attachedExcelBook. Both for amino acids and other kinds of building blocks one lettersymbols should be used.