Abstract In this study, we examined how branched-chain amino acids responded to varying experimental conditions through an adaptive principal component analysis (PCA) method. Three branched chain amino acids were chosen: L-leucine, L-valine and L-isoleucine and, four blends were created with varying ratios in order to analyze the effectiveness of this approach on multicomponent samples. Blended samples were created by dissolving the three individual amino acids in four different ratios: 1:1:1, 2:1:1, 1:2:1, and 1:1:2 and allowed to recrystallize with subsequent baking to remove moisture. X-ray Diffraction was performed on control samples to verify purity and determine crystal structure. Raman spectra was collected using a modular micro-Raman system and according to a randomized full-factorial design of experiment (DOE) with the factors including excitation laser power density, and stage temperature resulting in 180 unique spectral scans across the spectral range of 50−4000 cm−1. Assignment of each Raman band in the individual amino acid samples were compared to literature values. The adaptive PCA method was executed across each sample’s collective data set, and variance across each amino acid’s complete DOE was acquired. We identified the vibrational modes that responded to changes in the excitation laser power density setting and stage temperature setting in individual samples. The variance calculated using the adaptive PCA is illustrated using standard PCA biplots. The adaptive PCA method also revealed variance due to blend composition, where the vibrational modes in the blended samples were matched to those that appear in the individual amino acid Raman spectral data to calculate the wavenumber shift.

中文翻译：

---

使用具有自适应主成分分析的共焦拉曼光谱研究支链氨基酸菌株状态的显着特征

摘要 在这项研究中，我们通过自适应主成分分析 (PCA) 方法检查了支链氨基酸如何响应不同的实验条件。选择了三种支链氨基酸：L-亮氨酸、L-缬氨酸和 L-异亮氨酸，并创建了四种不同比例的混合物，以分析这种方法对多组分样品的有效性。通过以四种不同的比例溶解三种单独的氨基酸来创建混合样品：1:1:1、2:1:1、1:2:1 和 1:1:2，并允许在随后烘烤以去除水分时重结晶. 对对照样品进行 X 射线衍射以验证纯度并确定晶体结构。使用模块化微拉曼系统并根据随机全因子实验设计 (DOE) 收集拉曼光谱，其因素包括激发激光功率密度和载物台温度，从而在 50− 的光谱范围内进行 180 次独特的光谱扫描。 4000 厘米-1。将各个氨基酸样品中每个拉曼谱带的分配与文献值进行比较。在每个样本的集体数据集上执行自适应 PCA 方法，并获得每个氨基酸的完整 DOE 的方差。我们确定了响应单个样品中激发激光功率密度设置和载物台温度设置变化的振动模式。使用自适应 PCA 计算的方差使用标准 PCA 双标图进行说明。