A unique, semi-quantitative method has been developed to determine peptide lengths and their relative abundances (in mol% or ųg%) in (very) complex peptide mixtures like protein hydrolysates .In short, the method is based on algorithms applied to the quantitative data obtained from automated Edman degradation of the sample, usually for 16 cycles, and amino acid analysis (both of a hydrolyzed and a non-hydrolyzed sample) of the peptides and free amino acids extracted from the intact dairy products or other foods. The analysis results in a graphical presentation of the relative occurrence versus length of peptide.
Until now, we have used this method to compare different protein hydrolysates of unknown composition concerning the relative occurrence of peptides ranging in length from 1 - 15 amino acids (Mw range 0.1 - 1.5 kDa). It is possible to extend this region to peptides of 30-40 amino acid residues, although the reliability of the quantification may be somewhat lower. The presence of longer fragments, however, will be recognized/detected, and their molar percentage determined. This procedure has also been applied successfully to mixtures containing only 15% protein material (dry mass basis), and in the presence of sugars, lipids, fats, minerals, salts and vitamins.
Because the method and the algorithms applied contain certain assumptions (e.g. an equal efficiency of extraction for all fragments; contributions of Cys, Trp are not taken into account and N-terminally blocked fragments are of minor importance) it should be noted that, in our opinion, the method is suitable for comparison of different samples treated in the same way or to categorize them, more than that it pretends to give an exact reflection of the true situation.
To find out to what extend the results reflects the actual situation and for 'fine-tuning' the method, experiments with samples of known composition have been performed. The results of four such experiments on different mixtures of peptides and free amino acids (mix 1, 2, 3 and 4) are presented in the figure below.
As can be seen, there is a pretty good agreement between the experimental and theoretical values. The lengths of the peptides are given as 1-2, 2-3, 3-4 residues, etc., rather than simply 1,2, 3 etc. The reason is that peptides of length n will give a certain contribution to the determination of peptides with length n-1, but no significant contribution to the determination of peptides of length n+1 or n+2, etc., or to peptides of length n-2, n-3, etc, as can be seen in the Figure of Mix 4. This contribution, unfortunately, is variable (it depends on the specific amino acid sequence of the peptide, among other factors) and cannot be quantified exactly. Thus, a pentapeptide gives a 'false' signal at length '4' (4-5), as can be seen in the figure of Mix 4; addition of the two experimental values for peptides with length '4' (4-5) and '5' (5-6) in Mixture 4 gives a good correlation between the experimental and theoretical values.
Summarizing, the method is suitable for comparison of different samples treated in the same way or to categorize them. Experiments with samples of known composition indicate a good correlation between theoretical and experimental relative values; however, one should be careful with absolute data deduced from the (semi-quantitative) relative values.
Furthermore, it should be noted that in case of the presence of S-S-linked peptides, the length of each individual polypeptide chain will be determined, as will be the case for most, otherwise linked, fragments.
Apart from giving information about the relative distribution profile; the amino acid analysis data also provides quantitative data about both the free amino acid content and the protein content. Therefore, this method is very suitable for comparative studies of protein hydrolysates, for example present in food or feed applications, especially for mixtures containing short peptides (< 15 amino acids). It does not suffer from the disadvantages of the GPC analysis or the copper-Sephadex method  , which yields only a very rough indication of the peptide profile.
Determination of peptide lengths and their relative abundance (in mol%) in four different mixtures of peptides of known
composition. The X-axis (peptide length) is divided into groups (1-2, 2-3, 3-4, etc. instead of 1, 2, 3 etc.) because a
peptide with length n adds a 'false' signal to the experimental determination (shaded bars) of peptides of length n-1 as
can be seen for Mix 4. The theoretical values (hatched bars) are plotted on the same axis, although they refer to peptide
lengths of exactly 1, 2, 3, etc1.
©Parts of the Figures and text have been published in Ref. 1.
 Siemensma, A.D., Weijer, W.J. and Bak, H.J. (1993) Trends in Food Sci. Technol. 4, 16-21
 Rothenbühler, E., Waibel, R. and Solms, J. (1979) Anal. Biochem. 97, 367-375.