Protein-rich microalgal biomass compositions of optimized sensory quality

Abstract

The invention relates to a method for determining the organoleptic quality of a protein-rich microalgal biomass composition, comprising the determination of the content of 11 volatile organic compounds, wherein the 11 volatile organic compounds are pentanal, hexanal, 1-octen-3-ol, 2-pentylfuran, octanal, 3,5-octadien-2-ol (or 3-octen-2-one), 3,5-octadien-2-one, nonanal, 2-no-nenal, (E,E)-2,4-nonadienal and hexanoic acid.

Claims

1. A method for determining the organoleptic quality of a protein-rich microalgal biomass composition, comprising determining the content of 11 volatile organic compounds, the 11 volatile organic compounds being pentanal, hexanal, 1-octen-3-ol, 2-pentylfuran, octanal, 3,5-octadien-2-ol (or 3-octen-2-one), 3,5-octadien-2-one, nonanal, 2-nonenal, (E,E)-2,4-nonadienal and hexanoic acid, characterized in that the microalgal biomass comprises more than 50% proteins by dry weight of biomass and in that the microalgae are of the Chlorella genus.

2. The method as claimed in claim 1, wherein the content of 11 volatile organic compounds is determined by SPME/GC of by SPME/GC-MS.

3. The method as claimed in claim 2, characterized in that the content of 11 volatile organic compounds is determined by the surface area of the chromatography peaks after SPME/GC.

4. The method as claimed in claim 1, characterized in that the content of 11 volatile organic compounds is compared to that of a reference protein-rich microalgal biomass or biomasses for which the organoleptic qualities are defined.

5. The method as claimed in claim 4, wherein the content of the 11 volatile organic compounds is determined by the surface area of the chromatography peaks corresponding to the 11 volatile organic compounds and compared to the content of the 11 volatile organic compounds in the reference protein-rich microalgal biomass or biomasses for which the organoleptic qualities are defined.

6. The method as claimed in claim 1, characterized in that the microalgae are selected from the group consisting of Chlorella vulgaris, Chlorella sorokiniana and Chlorella protothecoides.

7. The method as claimed in claim 6, wherein the microalgae is Chlorella protothecoides.

8. A method for defining an analytical profile of volatile organic compounds making it possible to evaluate the organoleptic quality of the protein-rich microalgal biomass compositions, comprising: the construction of a first matrix associating microalgal biomass compositions, including two controls of acceptable and unacceptable organoleptic quality, with the evaluation of their organoleptic qualities by a sensory panel of at least 15 individuals, the construction of a second matrix associating with these same compositions their characterization by a volatile organic compound analysis profile, and the correlation of the first matrix with the second to produce a relationship model on the basis of which the compositions having an optimized organoleptic profile can thus be characterized by their analytical profile of volatile organic compounds; characterized in that the microalgal biomass comprises more than 50% proteins by dry weight of biomass and in that the microalgae are of the Chlorella genus.

9. The method as claimed in claim 8, characterized in that the descriptors of the sensory analysis comprise: the following odors: vegetable, mash, stock, rancid butter, cheese, manure, fermented, peanut and paint; and the following colors: yellow and green.

10. A method for determining the organoleptic quality of a protein-rich microalgal biomass composition, comprising determining the content of 4 volatile organic compounds, these 4 organic compounds being 3,5-octadien-2-ol (or 3-octen-2-one), 1-octen-3-ol, 3,5-octadien-2-one and (E,E)-2,4-nonadienal and calculating an overall flavor value from the sum of the individual flavor values of 3,5-octadien-2-ol (or 3-octen-2-one), 1-octen-3-ol, 3,5-octadien-2-one and (E,E)-2,4-nonadienal, characterized in that the microalgal biomass comprises more than 50% proteins by dry weight of biomass and in that the microalgae are of the Chlorella genus.

11. The method as claimed in claim 10, characterized in that the microalgae are selected from the group consisting of Chlorella vulgaris, Chlorella sorokiniana and Chlorella protothecoides.

12. The method as claimed in claim 11, wherein the microalgae is Chlorella protothecoides.

13. A method for selecting protein-rich microalgal biomass compositions having an acceptable organoleptic profile, characterized in that the organoleptic quality is determined by the method as claimed in 9, and that the composition is selected when the overall flavor value calculated by the method is between 0 and 40% relative to that of an organoleptically unacceptable reference microalgal biomass composition, characterized in that the microalgal biomass comprises more than 50% proteins by dry weight of biomass and in that the microalgae are of the Chlorella genus.

Description

DETAILED DESCRIPTION OF THE INVENTION

(1) For the purposes of the invention, a protein-rich microalgal biomass composition has an optimized sensory profile or an optimized organoleptic quality when its evaluation by a sensory panel concludes that there is an absence of off-notes which impair the organoleptic quality of said food formulations containing these microalgal biomass compositions.

(2) The term organoleptic quality is intended to mean the property of a food in terms of color and odor.

(3) These off-notes are associated with the presence of undesirable specific odorous and/or aromatic molecules which are characterized by a perception threshold corresponding to the minimum value of the sensory stimulus required to arouse a sensation.

(4) The optimized sensory profile or optimized organoleptic quality is then reflected by a sensory panel by obtaining the best scores on a scale of evaluation of the 2 sensory criteria (color and odors).

(5) The term approximately is intended to mean the value plus or minus 10% thereof, preferably plus or minus 5% thereof. For example, approximately 100 means between 90 and 110, preferably between 95 and 105.

(6) The term microalgal biomass composition is intended to mean a composition comprising at least 50%, 60%, 70%, 80% or 90% by dry weight of microalgal biomass. However, other ingredients can optionally be included in this composition.

(7) The term protein-rich is intended to mean a proteins content in the biomass of more than 50% by dry weight, preferably more than 55%, more preferably still more than 60%, 65% and 70% by dry weight of biomass.

(8) For the purposes of the present invention, the term microalgal biomass should be understood in its broadest interpretation and as denoting, for example, a composition comprising a plurality of particles of microalgal biomass. The microalgal biomass is derived from whole microalgal cells.

(9) A certain number of prior art documents, such as international patent application WO 2010/045368, describe methods for the production and use in food of protein-rich Chlorella microalgal biomass.

(10) The microalgae in question in the present invention are therefore microalgae of the Chlorella genus, more particularly Chlorella protothecoides, more particularly still Chlorella deprived of chlorophyll pigments, by any method known per se to those skilled in the art (either because the culture is carried out in the dark, or because the strain has been mutated so as to no longer produce these pigments).

(11) In particular, the microalgae can be chosen, non-exhaustively, from Chlorella protothecoides, Chlorella kessleri, Chlorella minutissima, Chlorella sp., Chlorella sorokiniana, Chlorella luteoviridis, Chlorella vulgaris, Chlorella reisiglii, Chlorella ellipsoidea, Chlorella saccarophila, Parachiorella kessleri, Parachiorella Prototheca stagnora and Prototheca moriformis. Thus, in one quite particular embodiment, the microalgal biomass composition is a Chlorella biomass composition, and in particular a Chlorella protothecoides biomass composition.

(12) The fermentative process described in this patent application WO 2010/045368 thus allows the production of a certain number of microalgal biomass compositions of variable sensory quality.

(13) The method as described in the present document therefore makes it possible to select the protein-rich microalgal biomass compositions which have an acceptable organoleptic profile, especially for food applications, without having to organize organoleptic evaluations by a panel of individuals in order to do so.

(14) 1. Definition of the Sensory Profile by Detecting 11 Volatile Organic Compounds

(15) The applicant company has discovered that the sensory profile of a protein-rich microalgal biomass composition can be defined by the nature and the threshold of detection of odorous specific molecules, in particular of specific volatile organic compounds.

(16) Indeed, it has identified 11 volatile organic compounds, the content of which in a protein-rich microalgal biomass composition makes it possible to determine the organoleptic quality of said composition.

(17) These 11 volatile organic compounds are the following: pentanal, hexanal, 1-octen-3-ol, 2-pentylfuran, octanal, 3,5-octadien-2-ol (or 3-octen-2-one), 3,5-octadien-2-one, nonanal, 2-nonenal, (E,E)-2,4-nonadienal and hexanoic acid.

(18) Thus, the present invention relates to a method for determining the organoleptic quality of a protein-rich microalgal biomass composition, comprising determining the content of each of these 11 volatile organic compounds, the 11 volatile organic compounds being pentanal, hexanal, 1-octen-3-ol, 2-pentylfuran, octanal, 3,5-octadien-2-ol (or 3-octen-2-one), 3,5-octadien-2-one, nonanal, 2-nonenal, (E,E)-2,4-nonadienal and hexanoic acid.

(19) The method does not exclude determining the content of other volatile organic compounds. However, the 11 volatile organic compounds are sufficient to determine the organoleptic quality of a protein-rich microalgal biomass composition.

(20) Preferably, these volatile organic compounds are sampled by solid phase microextraction (SPME) and analyzed by gas chromatography GC, in particular by GC-MS (gas chromatography-mass spectrometry).

(21) The volatile fraction is extracted from the sample of the protein-rich microalgal biomass composition by heating said composition for a sufficient period of time in the presence of an SPME fiber.

(22) The fiber may, for example, be chosen, non-exhaustively, from the group consisting of carboxen and polydimethylsiloxane (CAR/PDMS), divinylbenzene, carboxen and polydimethylsiloxane (DVB/CAR/PDMS), an alloy of metal and of polydimethylsiloxane (PDMS), a Carbopack-Z fiber (graphitized carbon black), polyacrylate, Carbowax polyethylene glycol (PEG), and PDMS/DVB.

(23) Preferably, a DVB/CAR/PDMS fiber (df 50/30 m) is used.

(24) Here, the applicant company recommends using a wet extraction technique (aqueous suspension) between 40 and 70 C., preferably between 50 and 65 C., in particular approximately 60 C. for at least 10 minutes, preferably at least 15 minutes and for example between 15 minutes and 1 hour.

(25) Preferably, this extraction step is carried out in a sealed container. A sufficient amount of sample must be used, for example at least 1 g, especially between 1 g and 10 g and in particular approximately 2 g.

(26) These 2 g are then suspended in 7 ml water containing 1 g CaCl.sub.2, 200 l HCl and 2.32 g hexanal-d12 (internal standard), placed in a sealed SPME flask (20 ml).

(27) The volatile organic compounds are then desorbed at a temperature compatible with the type of SPME fiber used, for example between 220 and 250 C. for the fiber used in our tests, more precisely at 230 C., and injected into the analysis system.

(28) Preferably, the analysis is carried out by gas chromatography GC, in particular by GC-MS.

(29) Several GC/MS devices are commercially available, for example the GC/Mass Clarus spectrometer (PerkinElmer, USA), the Hewlett Packard 6890 gas chromatograph (Hewlett Packard, USA) and the Agilent 6890N gas chromatograph coupled to the Agilent 5973 selective mass detector.

(30) The ionization methods which can be used in GC/MS are for example mass spectrometry with electron impact ionization (EI), chemical impact ionization (CI), electrospray ionization, luminescent discharge, field desorption (FD), etc.

(31) The volatile substances extracted are more precisely desorbed here in the injector of the TSQ GC-MS system from Thermo Scientific, and then separated on a CPwax52 (60 m0.25 mm, 0.25 m) column with helium gas at 1.5 ml/min.

(32) The temperature program is: 50 C. isotherm for 3 min, then programming at 5 C./minute up to 230 C., then isotherm for 20 min.

(33) The detection is carried out by electron impact (EI) mass spectrometry (MS) and the compounds are identified by comparison with the EI spectra of the NIST library.

(34) Thus, the height or the surface area of the chromatography peak corresponding to the volatile organic compound correlates with the amount of said compound.

(35) The term surface area of the peak is intended to mean the surface area of a specific ion under the curve in the SPME-GC/MS chromatogram.

(36) Preferably, the content of one of the 11 volatile organic compounds is determined by the surface area of the peak of the specific ion of the SPME-GC/MS chromatogram corresponding to this volatile organic compound.

(37) The content of volatile organic compounds is determined, especially in comparison with that of a reference product.

(38) Thus, overall, a low total content of the 11 volatile organic compounds is associated with an optimized organoleptic quality. Conversely, a higher total content of the 11 volatile organic compounds is associated with a medium, or even poor or unacceptable, organoleptic quality.

(39) For example, the total content of a composition with an acceptable organoleptic quality is low when it is at least two times less than that of a composition with an unacceptable organoleptic quality, for example at least 2, 3 or 4 times less, and in a most demanding embodiment, at least 10 times less.

(40) 2. Definition of the Sensory Panel and Choice of Descriptors

(41) The present invention relates to a method for testing the organoleptic qualities of a protein-rich microalgal biomass composition comprising evaluation of the organoleptic qualities by a panel of testers. This evaluation can especially be carried out by the methods detailed below.

(42) The applicant company also provides a method for defining an analytical profile of volatile compounds making it possible to evaluate the organoleptic quality of the microalgal biomass compositions, comprising: the construction of a first matrix associating protein-rich microalgal biomass compositions, including preferably two controls of acceptable and unacceptable organoleptic quality, with the evaluation of their organoleptic qualities by a sensory panel of at least 15 individuals, the construction of a second matrix associating with these same compositions their characterization by a volatile organic compound analysis profile, and the correlation of the first matrix with the second to produce a relationship model on the basis of which the compositions having an optimized organoleptic profile can thus be characterized by their analytical profile of volatile organic compounds.

(43) A sensory panel is formed in order to evaluate the sensory properties of various batches of microalgal biomass compositions, in particular Chlorella protothecoides biomass compositions.

(44) In order to evaluate the sensory properties of the protein-rich microalgal biomass compositions, a panel of at least 15 individuals, for example 18 individuals, was brought together.

(45) This expert panel makes it possible to carry out analyses of the QDA (Quantitative Descriptive Analysis) type, conventionally referred to as sensory profiles (Stone, H., Sidel, J-L., Olivier, S., Woolsey, A., Singleton, R. C; (1974), Sensory evaluation by quantitative descriptive analysis, Food Technology, 28(11), 24-33).

(46) As clarified by standard NF ISO 11035: 1995, sessions for generating descriptors were undertaken in order to exhaustively describe the olfactory properties of the protein-rich microalgal biomass compositions.

(47) For this purpose, batches of protein-rich microalgal biomass compositions identified as being highly heterogeneous were placed in solution at 3% in water.

(48) Each panelist evaluates this solution in a closed glass jar which has been heated beforehand in a water bath to 55 C., and lists all the odors he or she senses from the product.

(49) During the sessions for generating descriptors, more than 60 terms were listed by the judges to describe the odor of the protein-rich microalgal biomass compositions.

(50) The list of descriptors was firstly reduced qualitatively (e.g: lawn odor=cut grass odor), in order to obtain a list of 16 descriptors, then some QDA sessions enabled the list to be further reduced (cf: ISO 4121:1987) to 9 sensory descriptors.

(51) Preferably, the reference products as presented in the following table are associated with each descriptor:

(52) TABLE-US-00001 Descriptor Reference vegetable Mixed herb at 3% mash Mashed potato flakes at 5.6% stock 1 KUB OR from the company MAGGI per 2 l water rancid butter Rancid butter at 2.5% cheese Gorgonzola rind at 2% manure Manure at 2% fermented Tryptone (yeast extract) at 0.75% peanut Ground peanuts at 1.5% paint Highly oxidized protein-rich microalgal composition at 3%

(53) Since the paint descriptor was the most organoleptically discriminating, it is recommended by the applicant company to use it as the main descriptor in order to establish the sensory classification of the various batches of protein-rich microalgal biomass compositions produced.

(54) Training the Panel

(55) Various exercises were carried out in order to train the panel in the use of intensity scales for each descriptor (NF ISO 08587:1992, ISO 08586-1:1993, ISO 08586-2: 1994).

(56) The performance of the panel was finally validated by carrying out a profile exercise 3 times with the same batches of protein-rich microalgal biomass compositions; since the panel was considered to be discriminating, consensual and reproducible (method described in: Pages, J., L, S., Husson, F., Une approche statistiques de la performance en analyse sensorielle descriptive [A statistical approach to performance in descriptive sensory analysis], Sciences des aliments, 26(2006), 446-469), the tool can be used for the sensory analysis of the various batches of protein-rich microalgal biomass compositions, using the QDA method.

(57) Sensory Profile

(58) The panel analyses each protein-rich microalgal biomass composition one after the other on intensity scales for each descriptor.

(59) The questionnaire for one profile session (for 1 sample) is as follows:

(60) TABLE-US-00002 Color light dark Yellow Green Odors not quite perceived weak quite weak medium strong strong 0 1 2 3 4 5 vegetable mash stock rancid butter cheese manure fermented peanut paint

(61) Analyses of variance (ANOVAs) are carried out in order to evaluate the discriminating capacity of the descriptors (descriptors of which the p-value associated with the Fisher test is less than 0.20 for the Composition effect in the model DescriptorComposition+Panelist).

(62) The Composition effect is interpreted as the discriminating capacity of the descriptors: if there is no effect (Critical Probability>0.20), the various batches of compositions were not discriminated according to this criterion.

(63) The smaller the critical probability, the more discriminating the descriptor.

(64) The paint descriptor stood out as one of the most significant descriptors for characterizing the acceptability of a batch; the grade obtained for this descriptor will serve for sensory classification.

(65) This classification therefore then serves as a basis for studying the analytical profile of the volatile organic compounds and selecting molecules responsible for the poor organoleptic quality of the microalgal biomass compositions.

(66) Thus, the profile of the volatile organic compounds of the microalgal biomass compositions is determined. It is determined by any method known to those skilled in the art, and preferably by SPME/GC-MS, as detailed above.

(67) The analysis of the volatile compounds gives very complex GC-MS chromatograms, with a very large number of peaks. By means of analyses of variance and linear regressions, the volatile organic compounds which correlate best with the results obtained for the sensory matrix and with the paint odor.

(68) Thus, an optimized organoleptic profile is associated and characterized by an analytical profile of volatile organic compounds.

(69) In one preferred embodiment, the various organic compounds selected will be considered in terms of their total content, in comparison with reference compositions, especially as defined above. In particular, the total surface area of the chromatography peaks corresponding to the volatile organic compounds selected will be considered and compared.

(70) 3. Simplified Model Based on Four Volatile Organic Compounds Having an Impact on the Overall Odor

(71) In this preferred embodiment, the applicant company found that it is advantageously possible to establish an overall flavor value for the protein-rich microalgal biomass compositions having an optimized sensory profile, which overall value is based on 4 volatile organic compounds chosen from the 11 organic compounds identified above.

(72) These volatile organic compounds are selected on the basis of their criterion of low olfactory threshold. The overall flavor value is then established according the relationship:

(73) Overall flavor value= of the individual flavor values of 3,5-octadien-2-ol (or 3-octen-2-one), 1-octen-3-ol, 3,5-octadien-2-one and (E,E)-2,4-nonadienal.

(74) Total FV=FV(3,5-octadien-2-ol), FV(1-octen-3-ol), FV(3,5-octadien-2-one), and FV[(E,E)-2,4-nonadienal],

(75) where FV=Concentration of the compound x/olfactory threshold of the compound x

(76) As will be shown in the examples below, the protein-rich microalgal biomass compositions having a low overall flavor value of between 0 and 40%, relative to that of an organoleptically unacceptable reference microalgal biomass composition, are certain to have an optimized sensory profile.

(77) The invention will be understood more clearly from the examples which follow, which are intended to be illustrative and nonlimiting.

EXAMPLES

Example 1. Definition of the Sensory Test

(78) The perception of the protein-rich microalgal biomass composition is determined by solubilization in water, the neutral medium par excellence.

(79) A sensory panel was therefore formed to evaluate, according to the methodology set out above, the sensory properties of various batches of biomass of protein-rich microalgae, prepared according to the teaching of patent application WO 2010/045368.

(80) 18 batches of microalgal biomass were tested: batch 11, batch 12, batch 14, batch 33, batch 34, batch 42, batch 43, batch 44, batch 54, batch 81, batch 82, batch 83, batch 84, batch 85, batch 92, batch 93, batch 111, batch 112.

(81) The result of such characterization of the batches is given based on the highly characteristic descriptor of the odor of paint.

(82) Data Processing:

(83) The analyses were carried out using the R software (freely sold):

(84) R version 2.14.1 (Dec. 22, 2011)

(85) Copyright (C) 2011 The R Foundation for Statistical Computing

(86) ISBN 3-900051-07-0

(87) Platform: i386-pc-mingw32/i386 (32-bit)

(88) The software is a working environment which requires the loading of modules containing the calculation functions.

(89) The modules used for the processing of profile data are as follows: For the ANOVA: Package car version 2.0-12 For the Linear Regression: Package stats version 2.14-1

(90) The ANOVA shows significantly different results from one product to the next:

(91) TABLE-US-00003 Anova table (Type-III tests) Response: Data [,paint] Sum of squared deviations from the mean df F value Pr(>F) (mean) 327.94 1 234.1068 <2.2 10.sup.16 Composition 684.40 17 28.7393 <2.2 10.sup.16 Panelist 118.84 19 4.4649 7.097 10.sup.09 Residues 410.44 293

(92) The mean values obtained, by product, are as follows:

(93) TABLE-US-00004 standard Compositions mean deviation repetition Batch 092 0.00 0.00 8 Batch 111 0.00 0.00 9 Batch 12 0.10 0.10 10 Batch 112 0.28 0.12 36 Batch 43 0.29 0.17 21 Batch 44 0.33 0.26 12 Batch 33 0.36 0.36 11 Batch 11 0.40 0.40 10 Batch 81 0.96 0.25 23 Batch 14 1.00 0.50 9 Batch 82 1.50 0.34 24 Batch 34 1.64 0.38 22 Batch 42 1.96 0.25 49 Batch 93 2.67 0.58 9 Batch 84 3.23 0.36 22 Batch 54 4.09 0.41 11 Batch 83 4.24 0.16 34 Batch 85 4.60 0.22 10

(94) FIG. 1 gives the classification of the various batches in light of the grade given by the panelists based on this paint criterion.

(95) The classification is thus as follows, in increasing order of paint grade:

(96) batch 92>batch 111>batch 12>batch 112>batch 43>batch 44>batch 33>batch 11>batch 81>batch 14>batch 82>batch 34>batch 42>batch 93>batch 84>batch 54>batch 83>batch 85.

(97) Batch 92 is therefore defined as the control for acceptable organoleptic quality for this paint descriptor.

(98) Batch 85, for its part, is defined as the control for unacceptable organoleptic quality for this paint descriptor.

(99) This organoleptic classification having now been established, it is possible, efficiently according to the invention, to analyze the SPME/GC-MS profile of these samples in order to identify the reference molecular targets that will make it possible to define the quality of the compositions produced.

Example 2. Identification of the Volatile Organic Compounds (VOCs), by SPME/GC-MS, Associated with Unacceptable Paint Odor Organoleptic Classifications

(100) In order to carry out the SPME/GC-MS analysis of the 18 various batches of microalgal biomass compositions, the process is carried out as indicated above in aqueous suspension.

(101) Analysis of the Volatile Compounds on Products in Aqueous Suspension

(102) The volatile compounds were analysed in aqueous suspension in order to reduce the matrix effect, and an internal standard was added.

(103) Visually, as shown in FIG. 2, the GC-MS chromatograms remain very complex, with a very high number of compounds.

(104) The first approach consists in comparing the chromatographic profiles, in integrating all the peaks between 3.2 and 35.0 min (TIC, total ion current), and in checking whether these untreated results enable a link to be made to the sensory classification.

(105) The comparison of the chromatographic profiles and the integration of all the peaks between 3.2 and 35.0 min (TIC, total ion current)see FIG. 3do not enable a link to be made to the sensory classification.

(106) Because of the high complexity of the chromatograms, it is difficult to visually distinguish acceptable products from unacceptable products.

(107) The integration of the surface areas of the chromatograms also does not enable a clear distinction to be made between acceptable and unacceptable products.

(108) Moreover, this approach with untreated data does not enable it to be known which volatile compound(s) is (are) responsible for the off-notes or undesirable tastes or odors, nor to specifically monitor their appearance, nor to have any information on how they are formed.

(109) A second approach consisted in adding to the list of volatile organic compounds of the above model by listing the volatile compounds identified on the SPME/GC-MS chromatograms which appear to accompany the organoleptic classifications.

(110) From the GC-MS-olfactometric analysis of six samples, certain volatile compounds stood out; predominantly aldehydes derived from the degradation of the lipid fraction of the protein-rich microalgal biomass compositions, which are apparently responsible for the off-notes or undesirable tastes or odors.

(111) In this second approach, it was thus decided to monitor some of these molecules selected by GC-MS-olfactometry and GC-MS of the various products.

(112) In order to select the representative volatile organic compounds, a series of analyses of variance is carried out so as to keep only the volatile organic compounds which actually differ from one composition to the other given the variability of the SPME-GC/MS measurement.

(113) The model is the following: Volatile organic compoundComposition; only the compounds for which the critical probability associated with the Fisher test is less than 0.05 are retained.

(114) Two examples of ANOVA on the compounds 2-nonenal and 3,5-octadien-2-one are given here:

(115) TABLE-US-00005 Anova table (Type-III tests) Sum of squared deviations from the mean df F value Pr(>F) 2-nonenal (mean) 292.13 1 72.8516 3.51 10.sup.06 Composition 664.79 17 9.7522 0.0002394 Residues 44.11 11 3,5-Octadien-2-one (peak 2) (mean) 56344 1 20.0633 0.0009326 Composition 72570 17 1.5201 0.2424099 Residues 30891 11

(116) It appears on the first volatile organic compound (2-nonenal) that the composition effect is significant (critical probability <0.00025), which means that there is a significant difference between the products given the variability of the measurement.

(117) On the second compound (3,5-octadien-2-one, peak 2), the composition effect is not significant (critical probability >0.05). Thus, for the study, 2-nonenal will be retained but 3,5-octadien-2-one will not.

(118) After this first selection of volatile compounds, linear regression models are established: this involves explaining the paint variable by each compound one by one.

(119) As many models as there are compounds are therefore constructed. The model is the following: PaintCompound.

(120) In order to select the final list of compounds identified as responsible for the unacceptable organoleptic classifications (off-notes) observed, only the compounds for which the critical probability associated with Student's test is less than 0.05 (test for nullity of the linear regression coefficient) will be retained.

(121) The R.sup.2 associated with the model is an indicator for quantifying the percentage of variability explained by the compound. It may not be very high, but significant; for this reason, it is chosen to select the compounds according to the critical probability (so as not to neglect a compound which has little but significant influence on the paint odor described by the panel).

(122) Coefficients:

(123) TABLE-US-00006 Estimator Std value t value Pr(>|t|) (ordinate at 0.669037 0.138335 4.836 0.000182 the origin) Hexanal 0.019196 0.002593 7.404 1.49 10.sup.06 Residual standard error: 0.4444 on 16 degrees of freedom Multiple R.sup.2: 0.7741, adjusted R.sup.2: 0.76 Statistic F: 54.82 on 1 and 16 degrees of freedom, critical probability: 1.491 10.sup.06

(124) Coefficients:

(125) TABLE-US-00007 Estimator Std value t value Pr(>|t|) (ordinate at 0.45338 0.24852 1.824 0.0868 the origin) 3,5-octadien- 0.04346 0.01614 2.693 0.0160 2-one (peak 1) Residual standard error: 0.7756 on 16 degrees of freedom Multiple R.sup.2: 0.3119, adjusted R.sup.2: 0.2689 Statistic F: 7.252 on 1 and 16 degrees of freedom, critical probability: 0.016

(126) For these 2 compounds, hexanal and 3,5-octadien-2-one (peak 1), the critical probability is lower than 0.05, therefore they are correlated with the paint odor described by the panel.

(127) The 11 compounds of the study are found to be well correlated with the paint descriptor.

(128) These 11 molecules selected are listed in the table below:

(129) TABLE-US-00008 Reten- Spe- tion Olfactory cific time threshold ion Molecule (min) Odor (ppb) m/z 1 pentanal 6.24 Green 18* 44 2 Hexanal 8.24 Cut grass, green 4.5 82 apple 3 1-octen-3-ol 17.99 Mushroom-solvent 1/0.05* 57 (paint), mushroom- ink 4 2-pentylfuran 12.10 floral 6 81 5 Octanal 13.82 Floral-citrus 0.7 84 6 3,5-octadien-2-ol 17.03 Floral-zest .sup.0.1** 111 or 3-octen-2-one 7 3,5-octadien-2- 19.96 + floral .sup.0.1** 95 one (2 peaks) 21.24 8 Nonanal 16.68 Floral-green, 1 57 floral 9 2-Nonenal 20.37 Vegetable, oil 0.08 83 10 (E,E)-2,4- 24.42 Oily-oxidized 0.09 81 nonadienal 11 Hexanoic acid 27.51 Cheese, rancid 3000 60 *olfactory threshold according to H. Jelen, Journal of Chromatographic Science, vol. 44, August 2006 **estimated value Unless indicated otherwise, the olfactory threshold is taken from www.leffingwell.com/odorthre.htm

(130) As shown in FIG. 4, the unacceptable products appear to be much more loaded with these 11 volatile compounds than the acceptable samples.

(131) The statistical analysis confirms that all 11 molecules are significant (except the second peak of 3,5-octadien-2-one at 21.24 min).

(132) In conclusion, monitoring these 11 molecules (pentanal, hexanal, 1-octen-3-ol, 2-pentylfuran, octanal, 3,5-octadien-2-ol (or 3-octen-2-one), 3,5-octadien-2-one (first of the 2 peaks), nonanal, 2-nonenal, (E,E)-2,4-nonadienal, hexanoic acid) makes it possible to distinguish acceptable products from unacceptable products on the basis of the paint criterion, by analyzing the volatile substances of the product placed in aqueous suspension.

(133) Creating the Simplified Model

(134) In order to simplify the model, it is decided to retain the compounds having the greatest impact on the overall odor of the protein-rich microalgal biomass compositions according to the invention, that is to say the compounds with extremely low olfactory thresholds.

(135) These individual flavor values (=concentration of the compound/olfactory threshold thereof) are represented in FIG. 5.

(136) Taking into account the concentration and the olfactory threshold of each compound, four compounds appear to be particularly important for the sensory properties of the protein-rich microalgal biomass compositions according to the invention: 3,5-octadien-2-ol or 3-octen-2-one (floral-zest), 1-octen-3-ol (mushroom-solvent, paint, mushroom-ink), 3,5-octadien-2-one (the first of the two peaks, floral), and (E,E)-2,4-nonadienal (oily-oxidized).

(137) The individual descriptors of these four compounds bring together very well the overall perceived odor of the unacceptable protein-rich microalgal biomass compositions according to the invention.

(138) It is therefore possible to establish an overall flavor value for the protein-rich microalgal biomass compositions based on these four compounds:

(139) Overall flavor value= of the individual flavor values of 3,5-octadien-2-ol (or 3-octen-2-one), 1-octen-3-ol, 3,5-octadien-2-one and (E,E)-2,4-nonadienal.

(140) As shown in FIG. 6, it is henceforth easy to classify the various batches of protein-rich microalgal biomass compositions into two families: acceptable: these are batches 111, 92, 12, 112, 43, 33, 33, 81, 14, 44, 82 and 34; unacceptable: these are batches 83, 84 and 85.

(141) It should be noted that batch 85, which is a batch of organoleptically unacceptable quality according to example 1, has a flavor value of 100%.

(142) The acceptable batches therefore do indeed have an overall flavor value of between 0 and 40% compared to that of an unacceptable reference microalgal flour composition, in this case batch 85.

(143) Batches 42, 93 and 54 have an overall flavor value, based on the four organoleptic compounds, far greater than that of the reference batch 85.

(144) However, it should be noted that batch 85 was defined as unacceptable on the basis solely of the paint descriptor.

(145) Batches 42, 93 and 54, in terms of the analysis of volatile organic compounds, are particularly affected, undoubtedly by a synergistic effect between volatile organic compounds.

(146) This does not detract from the fact that the simplified model based on this selection of the 4 volatile organic compounds from the reference 11 makes it possible to classify the protein-rich microalgal biomass compositions into two distinct and easily identifiable families.

DESCRIPTION OF THE FIGURES

(147) FIG. 1: Average grade obtained for the paint descriptor for each of the compositions

(148) FIG. 2: Chromatograms (TIC) of the volatile organic compounds taken from samples by SPME in aqueous suspension

(149) FIG. 3: Integration of all the peaks for the zone 3.2-35.0 min of the chromatograms (SPME in aqueous suspension)

(150) FIG. 4: Relative contents of 11 selected compounds taken from the sample space by SPME in aqueous suspension

(151) FIG. 5: Individual flavor values for the 11 compounds selected

(152) FIG. 6: Overall flavor value based on 4 compounds