Microbiota by HOMIM Microarray 12 Sample swabs from cheeks, tongue, and alveolar ridges (牙槽嵴) Exclude 4 low-yield samples DNA extraction C-section: n=38: Vaginally delivered n=25 Human Oral Microbe Identification Microarray HOMIM HOM|M,2008) . Simultaneous detection of about 270 of the most prevalent, cultivated and not-yet-cultivated oral bacterial species Targeting 16SrRNA .Human Oral Microbe Identification using Next Generation Sequencing(HoMINGS, 2014)
Microbiota by HOMIM Microarray Sample Swabs from cheeks, tongue, and alveolar ridges(牙槽嵴) DNA extraction Exclude 4 low-yield samples C-section: n=38; Vaginally delivered: n=25. HOMIM Human Oral Microbe Identification Microarray( HOMIM, 2008) •Simultaneous detection of about 270 of the most prevalent, cultivated and not-yet-cultivated oral bacterial species; •Targeting 16SrRNA; •Human Oral Microbe Identification using Next Generation Sequencing(HOMINGS, 2014). 12
The Procedure of hemings ranes PCR amplification DNA Samples pooled sequenced using using V3-V4 16S Illumina MiSeg Isolation rDNA-specific then gel- extracted >50,000 reads per primers sample Quantitation Quantitation Quantitation Bioinformatic using sing using gPCR/ analysis using Nanodrop Nanodrop Bioanalyser Qiime/ProbeSeq
The Procedure of HOMINGS
Statistical Procedures 14 Body weight length: T test HOMIM microarray signals =0, if lack of signal =1, f signal≥1. Prevalence distribution between groups: x test, a=0.005 PLS-DA 样本点个数少于变量个数 ·自变量存在严重多重相关性的条件下进行回归建模 ·易于辨识系统信息与噪声
• Body weight & length: T test. • HOMIM microarray signals: • =0, if lack of signal; • =1, if signal ≥ 1. • Prevalence distribution between groups: χ 2 test, α=0.005. •PLS-DA • 样本点个数少于变量个数 • 自变量存在严重多重相关性的条件下进行回归建模 • 易于辨识系统信息与噪声 Statistical Procedures 14
Multivariate partial least-squares discriminant analysis, PLS-DA,15 偏最小二乘判别分析 速成 目的:区分样本来自哪个类别 方法: ·将响应变量/类别转化为哑变量[0,1] 11x1 ·偏最小二乘回归(2x22x2→01主成分分析+典型相关分析+多元线性回归 Xn1 Xn2 X: HOMIM信号、性别,身高体重>经孕周数、母乳喂养…Y生产方式 回归以及交叉验约吧 P(变量投影重要度 ·VIP=
速成: • 目的: 区分样本来自哪个类别 • 方法: • 将响应变量/类别转化为哑变量 𝟎, 𝟏 . • 偏最小二乘回归( 𝑥11 𝑥12 𝑥21 𝑥22 … 𝑥1𝑝 … 𝑥2𝑝 … … 𝑥𝑛1 𝑥𝑛2 … … … 𝑥𝑛𝑝 → 0,1 )(主成分分析+典型相关分析+多元线性回归) • X:HOMIM信号、性别、身高、体重、怀孕周数、母乳喂养……；Y: 生产方式. • 回归以及交叉验证的时候得到的量: R 2 , Q2 , VIP(变量投影重要度). • VIP= 𝑘 𝑅𝑑(𝑦;𝑡1,𝑡2…𝑡𝑚) ℎ=1 𝑚 𝑅𝑑(𝑦;𝑡ℎ) 𝑤ℎ𝑗 2 Multivariate partial least-squares discriminant analysis, PLS-DA, 偏最小二乘判别分析 15