Comprehensive sensory and chemical data on the flavor of 16 red wines from two varieties: Sensory descriptive analysis, HS-SPME-GC-MS volatile compounds quantitative analysis, and odor-active compounds identification by HS-SPME-GC-MS-O

This paper describes data collected on 2 sets of 8 French red wines from two grape varieties: Pinot Noir (PN) and Cabernet Franc (CF). It provides, for the 16 wines, (i) sensory descriptive data obtained with a trained panel, (ii) volatile organic compounds (VOC) quantification data obtained by Headspace Solid Phase Micro-Extraction – Gas Chromatography – Mass Spectrometry (HS-SPME-GC-MS) and (iii) odor-active compounds identification by Headspace Solid Phase Micro-Extraction – Gas Chromatography – Mass Spectrometry – Olfactometry (HS-SPME-GC-MS-O). The raw data are hosted on an open-access research data repository [1].


a b s t r a c t
This paper describes data collected on 2 sets of 8 French red wines from two grape varieties: Pinot Noir (PN) and Cabernet Franc (CF). It provides, for the 16 wines, (i) sensory descriptive data obtained with a trained panel, (ii) volatile organic compounds (VOC) quantification data obtained by Headspace Solid Phase Micro-Extraction e Gas Chromatography e Mass Spectrometry (HS-SPME-GC-MS) and (iii) odor-active compounds identification

Data
The dataset gathered, for the 16 wines from two grape varieties, 4 blocks of data: (1) the experimental factors (the grape variety, the vintage and the Protected Designation of Origin; Table 1), (2) the sensory descriptive data obtained with a trained panel using 33 sensory descriptors (Table 2), (3) the volatile organic compounds (VOC) quantification data obtained for 45 target odorants by Headspace Solid Phase Micro-Extraction e Gas Chromatography e Mass Spectrometry (Table 3) and (4)  Value of the data The data can help researchers to link sensory qualities of wines to their chemical composition [2]. The data can be used along with other datasets as a benchmark to develop methods and tools to predict the odor of wines [3].
The data can be compared to other wines varying in grape variety and vintage.
compounds, identified by Headspace Solid Phase Micro-Extraction e Gas Chromatography e Mass Spectrometry e Olfactometry (Table 4).

Wines
Two sets of French red wines from two grape varieties, 8 Pinot Noir wines (PN) and 8 Cabernet Franc wines (CF) were analyzed ( Table 1). The wines were selected out of 40 wines previously studied [4]. The main factors allowed for were vintage (2009 and 2010) and Protected Designation of Origin (PDO).

Sensory descriptive analysis
The sensory descriptive analysis of the 16 wines was performed at Groupe ESA, USC GRAPPE Senso'Veg (Angers, France).

Wines preparation
The wines were opened 30 minutes before the sensory evaluation and served (5 cL) in white ISO wine tasting glasses [5] at room temperature.

Sensory evaluation
Sixteen trained panelists, 6 women and 10 men (age range 35e71), participated in the sensory sessions. Sensory evaluation was performed according to recommended practices [6]. Before the sensory descriptive experiment, the judges were trained in 17 training sessions of 1-h each. This training consisted in a familiarization with the task and with the vocabulary and a selection of specific sensory descriptors for the wines set. During the familiarization step, the panelists did odor recognition tests on testing strip and on wines to become familiar with the sensory descriptors used for wines and smelled different standard odor references. These reference standards were adapted from [7]. During the sensory descriptors selection, the panelists were provided with an initial list of 84 descriptors. The list was elaborated by compiling terms from other lists employed in the description of wines from different varieties and geographical origins. Descriptors were arranged in the list by odor families: animal, burnt, floral, fruity, herbaceous, mineral, nut, spicy, undergrowth and others. Panelists modified the initial list of terms by removing those terms they considered irrelevant, ambiguous or redundant and by adding new attributes they considered pertinent while describing 15 wines of similar characteristics (grape variety and origin) as those considered in the present dataset. Finally, the terms cited by less than 15% of the panel were eliminated from the list. At the end of the training, the list included 33 descriptors (Table 2). During the sensory descriptive experiment, the judges had to evaluate monadically the 16 wines (orthonasal and retronasal olfaction) and to rate the intensity of 33 sensory descriptors on linear scales (14 cm); ratings were transformed into scores from 0 to 10. The protocol consisted in 3 repetitions by panelist for the orthonasal olfaction and 2 repetitions by panelist for the retronasal olfaction. Panelists Table 4 Linear Retention Index (apex) of odorant zones detected in GC-MS-O analysis of the wines, the name of the corresponding identified compounds and their CAS numbers. Compounds that appear in italics were tentatively identified owing to MS spectra, odor quality and LRI but available data could not allow discriminating between isomers. thus performed 5 evaluation sessions (one per week) and started with the 3 orthonasal sessions followed by the 2 retronasal sessions. The presentation order of the wines was counterbalanced according to a Williams Latin square. In order to depict the data collected through orthonasal olfaction, the rating scores were averaged over panelists and repetitions and then submitted to a standardized Principal Components Analysis (PCA) using the R software (version 3.4.0) and the FactoMineR package (version 1.34). The configuration of the 16 wines, as well as the correlations of the sensory descriptors with the two first principal components are shown in Fig. 1.

Volatile organic compounds quantitative analysis
The 16 wines were analyzed by GC-MS to quantify 45 target compounds (see Table 3). These analyses were carried out by a subcontracting external laboratory (ISO 9001 certification, afaq). The concentrations are reported in mg.L À1 in the headspace.
Extraction of volatile compounds was performed by Headspace Solid-Phase Micro-Extraction (HS-SPME) following an optimized protocol dedicated to wine volatile organic compounds used in routine by the specialized company. Wine samples were prepared by adding an internal standard, then acidified and salt saturated. A divinylbenzene (DVB)/carboxen (CAR)/polydimethylsiloxane (PDMS) SPME was used for headspace sampling. Extraction time was 60 min at 45 C. Volatile organic compounds analysis was then performed by GC-MS. The fiber was thermally desorbed in the 250 C splitless/split inlet of a GC (Shimadzu 2010) coupled with a mass spectrometer (Shimadzu QP2010þ).
Mass spectra were recorded in electron impact mode (70 eV) with a scan/SIM scanning method.
The identification of acetaldehyde, dimethyl sulfide, ethyl acetate, acetic acid, 2-ethylhexan-1-ol, propionic acid and phenol were carried out by comparison with reference mass spectra (WILEY257, NIST, in-house databases). Their quantification was based on an internal calibration by isotopic dilution with ethanal-13 C2, dimethyl sulfide-d 6 , ethyl acetate-13 C2, acetic acid-d 4 , 2-ethylhexan-1-ol-d 17 , propionic acid-d 5 and phenol-d 6 . The identification and quantification of all other compounds was based on a calibration method with these compounds as reference.
One randomly chosen wine sample was analyzed in five replicates in order to estimate the coefficient of variation on each compound, which ranged from 1.8% for phenol to 58.3% for 2-isobutyl-3-methoxypyrazine.

Analysis of wines by GC-MS-O
The 16 wines were analyzed by GC-MS-O at ONIRIS, UMR CNRS 6144 GEPEA Flavor group (Nantes, France).

Extraction methods
The wines were firstly oxygenated by a Venturi aerator, and then 7 mL of wine was poured in a 22 mL vial tightly capped with a Teflon/silicon septum. Volatile compounds from the wine samples were extracted by a representative procedure [8]. Prior to extraction, vials were incubated at 34 C for 1 h. After that, volatile compounds were extracted by Headspace Solid Phase Micro-Extraction (HS-SPME) with a Car/PDMS fiber (10 mm length, 85 mm film thickness; Supelco, Bellefonte, PA, USA) placed in the headspace of the vial for 10 min at 34 C.

Chromatographic conditions
The extracts were analyzed by GC (Agilent Technologies 6890N, Wilmington, DE, USA) coupled with a quadripole mass spectrometer (Agilent Technologies, 5973 Network), a FID and a sniffing port (ODP2, Gerstel, Baltimore, MD, USA) to identify odor-active compounds. Volatile compounds were desorbed in the injection port of the GC (T: 260 C; splitless mode for 5 min) and separated on a DB-Wax column (length: 30 m, internal diameter: 0.25 mm, film thickness: 0.5 mm). Hydrogen was used as carrier gas at constant flow (1 mL.min À1 ). The oven temperature program was set from 50 C (0 min) to 80 C at 5 C min À1 , from 80 C to 200 C at 10 C min À1 and from 200 to 240 C (4 min) at 20 C min À1 . Effluent from the end of the GC column was split 1:1:1 between the MS, the FID (250 C, air/H 2 flow: 450/40 mL.min À1 ), and the sniffing port. Peaks were integrated with MSD Chemstation software (Agilent Technologies). Mass spectra were recorded in electron impact mode (70 eV) between 33 and 300 m/z mass range at a scan rate of 2.7 scan s À1 .

Olfactometry
GC effluent was carried to the sniffing port using a deactivated and uncoated fused silica capillary column, heated to 200 C. The sniffing port was supplied with humidified air at 40 C with a flow of 600 mL.min À1 .
Olfactometry analyses were conducted by 8 experienced judges. Each judge performed one olfactometric analysis for each wine. Therefore, a total of eight GC-MS-O analyses were carried out for each wine. Judges were asked to express their perceptions via the olfactometric software interface [9], representing an aroma wheel made of 56 descriptors and designed for wine analysis. Characteristics of the perceptions were recorded throughout each judge's analysis and data were directly obtained from the olfactometric software. Odorant zones detected by at least 3 out of 8 judges were reported with their Linear Retention Index at apex and their associated odor descriptors.

Odorant compounds identification
The identification of compounds corresponding to each odorant zone was performed by comparing Linear Retention Index (LRI) and mass spectra of detected compounds with those of the databases (Wiley 6.0, NIST and in-house databases), by injection of the standard compounds when available, and by comparison of the odor perceived with those referenced in databases (in house database and The good scents company database [10]). The list of odor-active compounds is reported in Table 4. Compounds non-identified were named after their apex indices number.

Funding sources
These data were collected as part of the INNOVAROMA research program on wine, which was conducted with financial support from the regional councils of Pays de la Loire, Centre, Bretagne, and Bourgogne (DINOS-AGRALE-12-2011, DINOS-AGRALE-12-2012, PARI-DINOS-2013).