Raman spectra of blood obtained from 1–2-day-old chicksFigure 1A and B show the mean Raman spectra, along with one standard deviation, within the 1800–600 cm−1 region for blood obtained from 1–2-day-old male (n = 14) and female (n = 16) chicks. These blood samples comprise RBCs, white blood cells, platelets, and blood plasma. The spectra mainly exhibit mixed properties arising from bands attributed to porphyrin, a constituent of hemes in RBCs, and lipoproteins. The detailed explanations for the spectral properties of blood samples are discussed in SI1. General proteinaceous bands are evident at 1665 and 1003 cm−1, corresponding to amide III and the ring breathing mode of phenylalanine, respectively. In addition, bands associated with C–C, C=C, and C–H components in lipoproteins and porphyrin are observed at 1621, 1606, 1448, 1088, 938, and 856 cm−1. Characteristic bands originating from porphyrin are detected at 1580, 1546, 1338, 1305, 1224, 1215, 1128, 788, 754, and 672 cm−1. The detailed band assignments are summarized in Table 130,31,32,33,34.Figure 1Mean Raman spectra and one standard deviation (1SD) in the 1800–600 cm−1 region for blood from (A) male (n = 14) and (B) female (n = 16) chicks aged 1–2 days. The solid and dotted lines express the mean spectra and mean spectra ± 1SD, respectively. (C) Difference spectra calculated as the mean spectra of males minus that of females.Table 1 Band assignments and local coordinates for whole blood from young chickens aged 1 to 2 days recorded using 785-nm excitation30,31,32,33,34.The difference spectrum, calculated as the mean spectrum of females subtracted from that of males shown in Fig. 1A and B, showed peaks at 1639, 1570, 1225, 1193, 1032, 776, and 756 cm−1 in the positive domain and at 1687, 1544, 1301, 1089, 642 and 614 cm−1 in the negative domain (Fig. 1C). Gray-shaded areas denote wavenumber regions with significant differences in spectral intensities between males and females, as determined through the t-test (p < 0.05). Notable observations include a band at 1657 cm−1 in the 1700–1550 cm−1 region, peaks at 1301, 1193, 1089, and 1032 cm−1, and peaks in the 760–750 cm−1 region characteristic of the porphyrin (Fig. 1C).PCA analysis was conducted for these spectral data, including males and females. Figure 2A shows score plots of the PC 1 vs. PC 4, revealing a distinct classification of male and female datasets into two groups based on PC 4. In the loading plots of PC 1, PC 2, and PC 3, peaks corresponding to pyrroles were extracted. Conversely, in the loading plot of PC 4, peaks were extracted at 1687, 1615, 1546, 1302, 1225, 1192, 1032, and 751 cm−1, corresponding to wavenumber regions with significantly different spectral intensities between males and females, as defined using the t-test (Fig. 1C). Thus, distinguishing between male and female 1–2-day-old chicks based on the Raman spectra of blood is deemed feasible.Figure 2(A) Score plots of PCA (PC 1 vs. PC 4) derived from the Raman spectral dataset of blood from male and female chicks. (B) Loading plots of PC 1, PC 2, PC 3, and PC 4.Raman spectra of blood components: blood cells and blood plasma obtained from 1–2-day-old chicksTo delve into the specific blood component responsible for sex-dependent properties, separate Raman spectra of blood cells and plasma obtained through centrifugation were analyzed. Figure 3A and B depict the mean Raman spectra of blood cells for males (n = 12) and females (n = 11), respectively. Notably, both spectra show intense bands at 1621, 1549, 1225, 755, and 673 cm−1 attributed to the porphyrin, reflecting the spectral properties of RBCs. Wood et al. reported the Raman spectral differences between oxy- and deoxy-RBCs in human blood using a 785-nm excitation30. They concluded that the most intense band in deoxy-RBC spectra is observed at 1549 cm−1 rather than 1621 cm−1. The spectral patterns of blood cells from males and females shown in Fig. 3A and B reveal the characteristic pattern corresponding to oxy-RBCs.Figure 3Mean Raman spectra of RBCs from (A) male (n = 12) and (B) female (n = 11) chicks, and (C) blood plasma obtained from 1–2-day-old chickens. The solid and dotted lines express the mean spectra and mean spectra ± 1SD, respectively. (D) Difference spectra of RBCs calculated as the mean spectra of males minus that of females.Figure 3C shows the mean Raman spectra of blood plasma (n = 23), including both males and females, derived from the same blood samples used for blood cells in Fig. 3A and B. Intense bands observed at 1524 and 1157 cm−1 correspond to C=C and C–C stretching vibrational modes of carotenoids35,36,37, in addition to bands located at 1660, 1444, and 1004 cm−1 attributed to proteins (Table 1). The yellowish hue of blood plasma is attributed to carotenoids, such as lycopene and β-carotene, obtained from their dietary sources38. Figure 3D depicts the difference spectrum of blood cells, calculated as the mean spectrum of females subtracted from that of males. Wavenumber regions with significant differences in spectral intensities between males and females, as determined using the t-test (p < 0.05), are highlighted in gray. Numerous peaks are observed in both positive and negative directions, with a detailed discussion provided in subsequent sections.PCA was employed to analyze the Raman spectral dataset of blood cells to study the differences depending on chick sex. The PCA score plots (PC 1 vs. PC 3) revealed PC 3 as the discriminating factor between males and females (Fig. 4A). In the loading plot of PC 3, peaks were observed at 1546, 1526, 1305, 1214, 848, and 670 cm−1 in the positive domain and at 1638, 750, and 716 cm−1 in the negative domain (Fig. 4B). Notably, the difference spectrum of blood cells (Fig. 3D) exhibited a similar pattern to their loading spectrum (Fig. 4B), albeit with reversed positive and negative values. Thus, the loading plot effectively captured the spectral differences of RBCs depending on chick sex.Figure 4(A) Score plots of PCA (PC 1 vs. PC 3) calculated for the dataset of blood cells, including males and females. (B) Loading plots of PC 1, PC 2, and PC 3.Similarly, PCA was applied to the blood plasma dataset, including Raman spectra of males (n = 16) and females (n = 17). However, score plots for all PCs did not show separation based on sex-related spectral variations. Thus, the spectral disparities in blood depending on chick sex were primarily attributed to differences in RBCs.Raman spectra of oxy- and deoxy-RBCs obtained from 1–2-day-old chicksTo further investigate the origins of the differences in Raman spectra for RBCs between males and females, Raman spectra for oxy- and deoxy-RBCs of 1–2-day-old were obtained. Figure 5A shows Raman spectra within the 1800–600 cm−1 region for oxy- and deoxy-RBCs. A comparison of these two spectra revealed distinct differences in certain bands within the 1650–1500 cm−1 region depending on the states of oxy- or deoxy-RBCs. In the deoxy-RBC spectrum, as previously noted, the band intensity at 1544 cm−1 appeared stronger than that at 1619 cm−1. Conversely, in the oxy-RBC spectrum, the intensities of these two bands were almost comparable. Furthermore, the band intensities at 1636, 1579, and 1561 cm−1 were stronger in oxy-RBCs than in deoxy-RBCs, while a shoulder band observed at 1527 cm−1 was more intense in deoxy-RBCs. Subtraction spectrum, calculated as the spectrum of oxy-RBCs minus that of deoxy-RBCs, revealed that the more intense bands appeared in the spectra of oxy- and deoxy-RBCs in positive and negative domains, respectively (Fig. 5B). In the lower wavenumber region, the band at 1396 cm−1 was exclusively observed in the spectrum of oxy-RBCs, while the band at 787 cm−1 was more intense in the spectrum of deoxy-RBCs. Moreover, the relative band intensities at 1222 and 1213 cm−1, appearing as shoulder bands for each other, seemed to differ depending on the state of oxy- or deoxy-RBCs. The peaks at 1637, 1584, 1341, 1226, and 750 cm−1 in the positive domain of the loading plot of PC 3 (Fig. 3D) aligned well with the characteristic bands for oxy-RBCs shown in Fig. 5B, while the negative peaks at 1527 and 787 cm−1 in the loading plot of PC3 (Fig. 3D) corresponded to the bands for deoxy-RBCs shown in Fig. 5B. These results suggested that the PC 3 reflects the spectral differences attributed to the redox states of RBCs, with males having a higher proportion of oxy-RBCs.Figure 5(A) Mean Raman spectra of oxy- and deoxy-RBCs from 1–2-day-old chicks. (B) Subtraction spectra of (A) calculated as the spectrum of oxy-RBCs minus that of deoxy-RBCs.Quantitative analysis of Raman spectra to extract blood components with different concentrations depending on chick sexTo perform a quantitative analysis of the concentration differences based on the chick sex, PLS regression analysis was conducted. The concentration profile about arbitrary blood components defined as (male, female) = (2, 1) was given as an input parameter to conduct PLS regression analysis, and a calibration model reproducing the given concentration profile was built using the LOOC method29. Figure 6A and B depict the score and loading plots of PLS (Factors 1 and 2, respectively) generated for the dataset of whole blood. Factors 1 and 2 were identified as the main spectral components reflecting concentration profile. The R2 showed high accuracies for calibration and validation, with values of 1.0 and 0.84, respectively, by including up to six factors. In the loading plot of Factor 1, distinct bands were observed at 1636, 1567, and 1225 cm−1 in the positive domain and at 1540, 1521, and 1209 cm−1 in the negative domain, representing characteristic bands of oxy- and deoxy-RBCs, respectively, as shown in Fig. 5A. Conversely, in the loading plot of Factor 2, bands characteristic of deoxy-RBCs appeared at 1551 and 793 cm−1 in the negative domain. These results indicated that the spectral components derived from oxy- and deoxy-RBCs in blood were stronger and weaker, respectively, in males compared to females.Figure 6Score plots of PLS for Factors 1 vs. 2 calculated for the Raman spectral dataset of (A) whole blood and (C) blood cells. Loading plots of Factors 1 and 2, extracted as the spectral components with the concentration gradient given for (B) whole blood and (D) blood cells.Similarly, PLS analysis was performed on the dataset of Raman spectra for RBCs. Factors 1 and 2 of PLS were extracted as the spectral components corresponding to the given concentration gradient. The R2 value for validation and calibration reached 0.93 and 0.73, respectively, with up to three factors. While the accuracy in building the quantitative model for blood component concentrations appeared slightly reduced, it can be affirmed that the quantitative accuracy for biological samples was sufficiently maintained. The score and loading plots of Factors 1 and 2 are shown in Fig. 6C and D, respectively. In the loading plot of Factor 1, the spectral component for males exhibited stronger contributions to the positive bands at 1624, 1582, 1563, and 750 cm−1 than those for females. Focusing on the wavenumber region above 1500 cm−1, the band intensity at 1624 cm−1 was stronger than that at 1548 cm−1, indicating that Factor 1 likely reflected the Raman signal derived from oxy-RBCs. Conversely, in the loading plot of Factor 2, an opposite trend was observed in the wavenumber region above 1500 cm−1 compared to Factor 1; the band intensity at 1547 cm−1 was more intense than that at 1621 cm−1. These results suggest that Factor 2 expressed the contribution of deoxy-RBCs.In this way, a model for quantitative evaluation was constructed, assuming that one component was present in higher concentration in males than in females, to extract the spectral components contributing more precisely to chick sex discrimination in the samples of whole blood and RBCs. As results, it was reconfirmed that chick sexing using the Raman spectra of whole blood and RBCs was accomplished based on the characteristic Raman bands attributed to oxy- and/or deoxy-RBCs.
