A comprehensive R package for streamlining analytical chemistry workflows
labtools is specifically designed for our lab’s analytical chemistry workflows and daily data processing needs. This package prioritizes ease of use for researchers who may not be R experts over maximum flexibility.
Design Philosophy: - Laboratory-focused: Built around common analytical chemistry tasks and workflows - User-friendly: Simple function calls with sensible defaults for non-R experts - Workflow-oriented: Functions designed to work together in typical lab data processing pipelines - Trade-offs: Some flexibility is sacrificed for simplicity and ease of use
Please use according to your needs - if you require maximum flexibility, consider using the underlying packages directly (webchem, rcdk, etc.).
| Feature | Description |
|---|---|
| 🔍 Chemical Metadata Extraction | Enhanced automated retrieval from PubChem with robust error handling and CAS parsing |
| 🧬 Structure Database Management | Build and visualize chemical structure databases with advanced filtering |
| 📊 MS Data Processing | Export optimized databases for MS-FINDER and MS-DIAL software |
| 🔬 2D GC-MS Analysis | Process Canvas exports and combine multi-sample data with precision |
| ⚗️ Chemical Structure Conversion | SMILES to MOL file conversion with 2D coordinate generation |
| 📈 Semi-quantification Tools | Advanced analytical quantification with machine learning integration |
| 🎯 Interactive Visualization | Shiny-based chemical structure navigation and spectrum plotting |
💡 New to R? For detailed help on any function, use
?function_namein R console (e.g.,?extract_cid)
- Enhanced metadata extraction: Improved
parse_cas_clean(),extract_cid(), andextract_meta()functions - Better dependency management:
mspcompileris now optional - install only when needed - Robust error handling: Comprehensive input validation and informative error messages
- Improved documentation: Detailed examples and parameter explanations for all functions
- Enhanced synonyms extraction: Fixed and optimized PubChem synonyms retrieval
- ASCII compliance: All code now uses ASCII characters for better portability
- Comprehensive testing: All functions pass R CMD check with zero errors and warnings
# Install devtools if you haven't already
if (!requireNamespace("devtools", quietly = TRUE)) {
install.packages("devtools")
}
# Install labtools from GitHub
devtools::install_github("QizhiSu/labtools")- R ≥ 4.0.0
- Java ≥ 8 (for rcdk package)
- Internet connection (for PubChem API access)
Some functions require additional packages that are not automatically installed:
-
mspcompiler: Required for MSP library filtering (
filter_msp()function)# Install when needed devtools::install_github("QizhiSu/mspcompiler")
- Function help: Use
?function_name(e.g.,?extract_cid) - Package help: Use
help(package = "labtools") - Examples: Use
example(function_name)to run examples
library(labtools)
library(dplyr)
# 🔍 1. Extract chemical metadata from PubChem
data <- data.frame(
Name = c("Caffeine", "Aspirin", "Glucose"),
CAS = c("58-08-2", "50-78-2", "50-99-7")
)
# Extract CIDs and comprehensive metadata
data_with_cid <- extract_cid(data, name_col = "Name", cas_col = "CAS")
data_complete <- extract_meta(data_with_cid, cas = TRUE, flavornet = TRUE)
# 📊 2. Export for MS software
export4msdial(data_complete, polarity = "pos", output = "database_msdial.txt")
export4msfinder(data_complete, output = "database_msfinder.txt")
# ⚗️ 3. Convert SMILES to MOL files
smiles_data <- data.frame(
ID = c("Caffeine", "Aspirin"),
SMILES = c("Cn1cnc2c1c(=O)n(c(=O)n2C)C", "CC(=O)OC1=CC=CC=C1C(=O)O")
)
export_smiles_to_mol(smiles_data, output_dir = "mol_files")
# 🎯 4. Interactive chemical structure browser
navigate_chem(data_complete) # Opens Shiny app💡 Tip: Use ?extract_cid or
?export_smiles_to_mol to see detailed parameter
descriptions and more examples!
library(labtools)
# Extract CIDs from chemical identifiers
compounds <- data.frame(
Name = c("Caffeine", "Theobromine"),
CAS = c("58-08-2", "83-67-0")
)
# Extract CIDs with proper error handling
# Use ?extract_cid for full parameter details
compounds_cid <- extract_cid(
data = compounds, # Input data frame
name_col = "Name", # Column with chemical names (default: "Name")
cas_col = "CAS", # Column with CAS numbers (default: "CAS")
inchikey_col = "InChIKey", # Column with InChIKeys (default: "InChIKey")
verbose = TRUE # Show progress messages (default: TRUE)
)
# Extract comprehensive metadata
# Use ?extract_meta for all options
compounds_meta <- extract_meta(
data = compounds_cid, # Data frame with CID column (required)
cas = TRUE, # Extract CAS numbers (default: FALSE)
flavornet = TRUE, # Extract Flavornet data (default: FALSE)
synonyms = TRUE, # Extract synonyms (default: FALSE)
uses = TRUE, # Extract compound uses (default: FALSE)
verbose = TRUE # Show progress (default: TRUE)
)
# Add chemical classification using ClassyFire
# Use ?extract_classyfire for details
compounds_classified <- extract_classyfire(
data = compounds_meta, # Data frame with InChIKey column
inchikey_col = "InChIKey", # InChIKey column name (default)
name_col = "Name" # Name column for progress (default)
)Key Parameters Explained: - name_col, cas_col, inchikey_col:
Specify which columns contain identifiers (defaults: “Name”, “CAS”,
“InChIKey”) - verbose: Set to FALSE to suppress progress messages
(default: TRUE) - cas, flavornet, synonyms, uses: Enable
extraction of specific metadata types (all default: FALSE) - Functions
automatically handle network errors and provide informative progress
messages
# Export for MS-FINDER software
# Use ?export4msfinder for details
export4msfinder(
data = compounds_classified, # Data with required columns
output = "msfinder_database.txt" # Output file path (default)
)
# Export for MS-DIAL (positive mode)
# Use ?export4msdial for all options
export4msdial(
data = compounds_classified, # Data with required columns
polarity = "pos", # ESI polarity: "pos" or "neg" (default: "pos")
output = "msdial_pos.txt" # Output file path (default)
)
# Export for MS-DIAL (negative mode)
export4msdial(
data = compounds_classified,
polarity = "neg", # Negative mode for different adducts
output = "msdial_neg.txt"
)Required Columns: - MS-FINDER: Name, InChIKey, CID, ExactMass, Formula, SMILES - MS-DIAL: Name, ExactMass, SMILES, InChIKey
Polarity Options: - "pos": Generates [M+H]+, [M+Na]+, [M+K]+
adducts - "neg": Generates [M-H]-, [M+Cl]-, [M+HCOO]- adducts
# Prepare SMILES data
smiles_data <- data.frame(
Compound_ID = c("CAFF_001", "THEO_002"),
SMILES_String = c(
"Cn1cnc2c1c(=O)n(c(=O)n2C)C", # Caffeine
"Cn1cnc2c1c(=O)[nH]c(=O)n2C" # Theobromine
),
stringsAsFactors = FALSE
)
# Convert to MOL files with 2D coordinates
# Use ?export_smiles_to_mol for all parameters
result <- export_smiles_to_mol(
df = smiles_data, # Data frame with ID and SMILES columns
id_col = "Compound_ID", # Column with compound IDs (default: "ID")
smiles_col = "SMILES_String", # Column with SMILES strings (default: "SMILES")
output_dir = "molecular_structures" # Output directory (default: "mol_files")
)
# Check conversion results
print(result)
# Returns: list(success = 2, failed = 0, skipped = 0)Function Features: - Input validation: Checks for valid data frame and required columns - Error handling: Continues processing even if some SMILES fail - 2D coordinates: Automatically generates 2D coordinates for visualization - Progress tracking: Returns summary of successful/failed conversions - Robust processing: Handles NA/NULL values gracefully
# Process Canvas 2D GC-MS data with detailed parameters
# Use ?read_canvas for complete parameter documentation
canvas_data <- read_canvas(
path = "path/to/canvas/files", # Directory containing Canvas .txt files
ri_iden_tol = 30, # RI tolerance for identification (default: 30)
ri_p_iden_tol = 100, # Predicted RI tolerance for identification (default: 100)
ri_align_tol = 50, # RI tolerance for alignment (default: 50)
rt_2d_align_tol = 0.1, # 2D RT tolerance for alignment (default: 0.1)
keep = "area" # Data to keep: "area", "height", or "both" (default: "area")
)
# Normalize by internal standard (D8)
# Use ?normalize_area for details
normalized_data <- normalize_area(
df = canvas_data, # Data frame with peak area data
start_col = 12 # Starting column index of area data (default: 12)
)
# Filter areas based on sample codes
# Use ?keep_area for filtering options
filtered_data <- keep_area(
df = normalized_data, # Data frame with peak area data
sam_code = sample_codes, # Data frame mapping sample codes to names
start_col = 12, # Starting column index (default: 12)
keep_bk = TRUE, # Keep blank samples (default: TRUE)
keep_d8 = TRUE # Keep D8 internal standard (default: TRUE)
)Key Parameters Explained: - ri_iden_tol, ri_p_iden_tol:
Retention index tolerances for compound identification - ri_align_tol,
rt_2d_align_tol: Tolerances for aligning peaks across samples -
keep: Determines which peak data to retain (area is most common for
quantification) - start_col: Column index where peak area data begins
(typically after metadata columns) - keep_bk, keep_d8: Control
whether to retain blank samples and internal standards
Note on normalize_area(): This workflow uses normalize_area()
for QC monitoring/data exploration. For concentration quantification,
use calculate_con() instead (see the Semi-quantification Analysis
section for the correct workflow).
The semi-quantification workflow has two steps: select_std assigns the
best-matching standard to each detected compound via molecular
descriptor similarity, then calculate_con converts peak areas to
concentrations. Two calibration modes are available and auto-detected
from the data:
| Mode | When to use | Required std_md columns | Concentration formula |
|---|---|---|---|
| Calibration curve | Full multi-level calibration per standard | Slope, Intercept (+ optionally R2, Low_con, Low_area) |
(A / A_D8 − Intercept) / Slope |
| Single-point | One mixed standard at known concentration | STD_con, STD_area |
(A / A_D8 × STD_con) / STD_area |
For negative concentrations in calibration-curve mode, a single-point
fallback is applied: (A / A_D8 × Low_con) / Low_area.
Calibration-curve mode — standard data must contain Slope,
Intercept (required), and optionally R2, Low_con, Low_area:
std_md <- reference_standards # Name, Slope, Intercept, R2, Low_con, Low_area, descriptors...Single-point mode — standard data must contain STD_con (known
concentration, e.g. 10 µg/mL) and STD_area (measured peak area):
std_md <- allStd_md[allStd_md$Name %in% std$Name, ] |>
mutate(STD_area = std$STD[match(Name, std$Name)], # mean peak area from injections
STD_con = 10) |> # mix standard concentration (µg/mL)
relocate(STD_con, STD_area, .after = SMILES)Assigns the best-matching standard to each compound based on molecular
descriptor similarity. Mode and response column are auto-detected from
std_md columns; set std_res_col = FALSE to skip feature selection.
quan_res <- select_std(
std_md = std_md,
std_res_col = "Slope", # "STD_area" for single-point; omit = auto-detect
std_md1_col = 8, # first molecular descriptor column in std_md
data_md = target_compounds,
data_md1_col = 5, # first molecular descriptor column in data_md
top_npct_md = 30 # % top descriptors for RF feature selection
)sample_data <- left_join(sample_data,
quan_res[, c("Name", "Standard", "Similarity",
"Slope", "Intercept", "R2", "Low_con", "Low_area", # calibration mode
"STD_con", "STD_area")], # single-point mode
by = "Name"
)Converts peak areas to concentrations. Mode is auto-detected from the columns present.
concentrations <- calculate_con(
df = sample_data,
sam_weight = sample_weights, # NULL = instrument concentration only; see format below
start_col = 12 # first peak-area column
)Summarizes replicate injections as mean ± SD per sample name.
final_results <- organize_con(
df = concentrations,
sam_code = sample_codes, # see format below
start_col = 12,
digits = 3,
na2zero = TRUE,
bind_mean_sd = TRUE
)sam_weight — for concentration per sample weight (omit if
instrument concentration only):
| Column | Description |
|---|---|
Name |
Sample code, must match column names in peak-area data |
Weight |
Sample weight (e.g., in g) |
Volumn |
Final extract volume (e.g., in mL) |
Conversion: conc_in_sample = conc × Volumn / Weight (e.g., ng/g).
sample_weights <- data.frame(
Name = c("S1", "S2", "S3"),
Weight = c(0.50, 0.52, 0.48),
Volumn = c(1.0, 1.0, 1.0)
)sam_code — maps raw codes to display names for keep_area() and
organize_con():
| Column | Description |
|---|---|
Code |
Code appearing in column names or Comment column |
Name |
Display name; replicates sharing the same Name are grouped |
sample_codes <- data.frame(
Code = c("S1a", "S1b", "S2a", "S2b", "S3a"),
Name = c("S1", "S1", "S2", "S2", "S3")
)These functions are mutually exclusive — do not use them sequentially.
| Function | Formula | Output | Use case |
|---|---|---|---|
normalize_area() |
area / D8 |
Normalized peak area | QC monitoring, data exploration |
calculate_con() |
see table above | Concentration | Quantification with calibration |
Using both sequentially applies D8 normalization twice, giving incorrect results. Choose one based on your goal.
# Convert spectrum string to numeric vector
# Use ?update_spectrum for format details
spectrum_str <- "100:30 120:80 145:60 170:90 200:20"
spectrum <- update_spectrum(
spectrum_str = spectrum_str, # Spectrum in "mz:intensity mz:intensity" format
start_mz = 50, # Starting m/z value (default: 50)
end_mz = 500, # Ending m/z value (default: 500)
mz_step = 1, # m/z step size (default: 1)
digits = 0 # Decimal places for m/z rounding (default: 0)
)
# Create interactive spectrum plot
# Use ?plot_spectrum for plotting options
plot_spectrum(
spectrum = spectrum, # Named numeric vector (m/z as names, intensity as values)
range = 10, # Bin width for peak labeling (default: 10)
threshold = 5, # Minimum intensity (%) for peak labeling (default: 1)
max_ticks = 20 # Maximum number of x-axis ticks (default: 20)
)
# Calculate spectral similarity using cosine similarity
# Use ?cosine_similarity for similarity calculation details
spec1 <- update_spectrum("100:30 120:80 145:60")
spec2 <- update_spectrum("100:25 120:85 145:55")
similarity <- cosine_similarity(
spectrum1 = spec1, # First spectrum (named numeric vector)
spectrum2 = spec2 # Second spectrum (named numeric vector)
)
# Convert between MSP and data frame formats
# Use ?msp2df and ?df2msp for format conversion
msp_data <- msp2df(msp_file_path) # Convert MSP file to data frame
df2msp(df_data, "output.msp") # Convert data frame to MSP fileSpectral Data Formats: - Input: Spectrum strings in “mz:intensity mz:intensity” format - Processing: Converts to named numeric vectors for analysis - Output: Interactive plots and similarity scores
Key Functions: - update_spectrum(): Converts text format to
numeric vectors - plot_spectrum(): Creates interactive mass spectrum
plots - cosine_similarity(): Calculates spectral similarity (0-1
scale) - msp2df() / df2msp(): Convert between MSP files and data
frames
# Note: MSP filtering requires the mspcompiler package
# Install if needed: devtools::install_github("QizhiSu/mspcompiler")
# Prepare compound list for filtering
compounds_to_keep <- data.frame(
Name = c("Caffeine", "Aspirin", "Glucose"),
InChIKey = c("RYYVLZVUVIJVGH-UHFFFAOYSA-N",
"BSYNRYMUTXBXSQ-UHFFFAOYSA-N",
"WQZGKKKJIJFFOK-GASJEMHNSA-N"),
stringsAsFactors = FALSE
)
# Filter MSP spectral library based on compound list
# Use ?filter_msp for filtering options
filter_msp(
msp = "NIST_library.msp", # Path to the MSP library file
cmp_list = compounds_to_keep, # Data frame with compounds to keep (Name and InChIKey columns required)
keep_napd8 = TRUE, # Add Naphthalene-D8 to the filter list (default: TRUE)
output = "filtered_library.msp" # Output path for filtered MSP file
)
# Read MS-DIAL output files with filtering options
# Use ?read_msdial for reading options
msdial_data <- read_msdial(
file_path = "msdial_results.txt", # Path to MS-DIAL output file
keep_unknown = FALSE, # Keep unknown compounds (default: FALSE)
keep_spectrum = TRUE, # Keep spectrum data (default: FALSE)
keep_mean_sd = FALSE # Keep mean and SD columns (default: FALSE)
)MSP Library Management: - Purpose: Filter large spectral libraries to contain only compounds of interest - Requirements: Compound list must have Name and InChIKey columns - Features: Automatically includes Naphthalene-D8 internal standard - Output: Filtered MSP file ready for MS software
MS-DIAL Integration: - Input: MS-DIAL alignment result files - Options: Control retention of unknown compounds and spectral data - Output: Clean data frame ready for further analysis
# Step 1: Prepare data
chemicals <- data.frame(
Name = c("Benzene", "Toluene", "Xylene"),
CAS = c("71-43-2", "108-88-3", "1330-20-7")
)
# Step 2: Extract comprehensive metadata
chemicals_cid <- extract_cid(chemicals, name_col = "Name", cas_col = "CAS")
chemicals_meta <- extract_meta(chemicals_cid, cas = TRUE, synonyms = TRUE)
chemicals_class <- extract_classyfire(chemicals_meta)
# Step 3: Export for different platforms
export4msdial(chemicals_class, polarity = "pos", output = "pos_database.txt")
export4msdial(chemicals_class, polarity = "neg", output = "neg_database.txt")
export4msfinder(chemicals_class, output = "msfinder_database.txt")
# Step 4: Generate structure files
export_smiles_to_mol(chemicals_class, output_dir = "structures")# Process Canvas data
canvas_data <- read_canvas("canvas_files", keep = "area")
# Extract metadata
canvas_meta <- extract_cid(canvas_data, name_col = "Name", cas_col = "CAS")
canvas_complete <- extract_meta(canvas_meta, cas = TRUE)
# Normalize and analyze
canvas_normalized <- normalize_area(canvas_complete)
canvas_filtered <- keep_area(canvas_normalized, sample_codes)
# Semi-quantification
standards <- select_std(ref_standards, std_md1_col = 14, data_md = canvas_filtered,
data_md1_col = 5)
concentrations <- calculate_con(standards, sam_weight = sample_weights)
results <- organize_con(concentrations, sample_codes)专为分析化学工作流程设计的综合性R包
labtools 专门为分析化学实验室工作流程和日常数据处理需求而设计。本包优先考虑不太熟悉R语言的研究人员的易用性，而非最大的灵活性。
设计理念： - 实验室导向：围绕常见的分析化学任务和工作流程构建 - 用户友好：为非R专家提供简单的函数调用和合理的默认设置 - 工作流导向：函数设计为在典型的实验室数据处理流程中协同工作 - 权衡取舍：为了简单性和易用性，牺牲了一些灵活性
请根据您的需求使用 - 如果您需要最大的灵活性，请考虑直接使用底层包（webchem、rcdk等）。
| 功能 | 描述 |
|---|---|
| 🔍 化学元数据提取 | 从PubChem数据库自动检索，具有智能重试机制 |
| 🧬 结构数据库管理 | 构建和可视化化学结构数据库，支持高级过滤 |
| 📊 质谱数据处理 | 为MS-FINDER和MS-DIAL软件导出优化数据库 |
| 🔬 二维气相色谱-质谱分析 | 精确处理Canvas导出数据并合并多样本数据 |
| ⚗️ 化学结构转换 | SMILES到MOL文件转换，支持2D坐标生成 |
| 📈 半定量工具 | 集成机器学习的高级分析定量方法 |
| 🎯 交互式可视化 | 基于Shiny的化学结构导航和光谱绘图 |
💡 R语言新手？
对于任何函数的详细帮助，请在R控制台中使用 ?函数名
（例如：?extract_cid）
# 安装devtools（如果尚未安装）
if (!requireNamespace("devtools", quietly = TRUE)) {
install.packages("devtools")
}
# 从GitHub安装labtools
devtools::install_github("QizhiSu/labtools")- R ≥ 4.0.0
- Java ≥ 8 (rcdk包需要)
- 网络连接 (访问PubChem API)
- 函数帮助: 使用
?函数名(例如：?extract_cid) - 包帮助: 使用
help(package = "labtools") - 示例: 使用
example(函数名)运行示例
library(labtools)
library(dplyr)
# 🔍 1. 从PubChem提取化学元数据
data <- data.frame(
Name = c("咖啡因", "阿司匹林", "葡萄糖"),
CAS = c("58-08-2", "50-78-2", "50-99-7")
)
# 提取CID和综合元数据
data_with_cid <- extract_cid(data, name_col = "Name", cas_col = "CAS")
data_complete <- extract_meta(data_with_cid, cas = TRUE, flavornet = TRUE)
# 📊 2. 导出用于质谱软件
export4msdial(data_complete, polarity = "pos", output = "database_msdial.txt")
export4msfinder(data_complete, output = "database_msfinder.txt")
# ⚗️ 3. 转换SMILES为MOL文件
smiles_data <- data.frame(
ID = c("咖啡因", "阿司匹林"),
SMILES = c("Cn1cnc2c1c(=O)n(c(=O)n2C)C", "CC(=O)OC1=CC=CC=C1C(=O)O")
)
export_smiles_to_mol(smiles_data, output_dir = "mol_files")
# 🎯 4. 交互式化学结构浏览器
navigate_chem(data_complete) # 打开Shiny应用💡 提示: 使用 ?extract_cid 或
?export_smiles_to_mol 查看详细的参数说明和更多示例！
library(labtools)
# 从化学标识符提取CID
compounds <- data.frame(
Name = c("咖啡因", "可可碱"),
CAS = c("58-08-2", "83-67-0")
)
# 提取CID并进行适当的错误处理
# 使用 ?extract_cid 查看完整参数详情
compounds_cid <- extract_cid(
data = compounds, # 输入数据框
name_col = "Name", # 包含化学名称的列 (默认: "Name")
cas_col = "CAS", # 包含CAS号的列 (默认: "CAS")
inchikey_col = "InChIKey", # 包含InChIKey的列 (默认: "InChIKey")
verbose = TRUE # 显示进度信息 (默认: TRUE)
)
# 提取综合元数据
# 使用 ?extract_meta 查看所有选项
compounds_meta <- extract_meta(
data = compounds_cid, # 包含CID列的数据框 (必需)
cas = TRUE, # 提取CAS号 (默认: FALSE)
flavornet = TRUE, # 提取Flavornet数据 (默认: FALSE)
synonyms = TRUE, # 提取同义词 (默认: FALSE)
uses = TRUE, # 提取化合物用途 (默认: FALSE)
verbose = TRUE # 显示进度 (默认: TRUE)
)
# 使用ClassyFire添加化学分类
# 使用 ?extract_classyfire 查看详情
compounds_classified <- extract_classyfire(
data = compounds_meta, # 包含InChIKey列的数据框
inchikey_col = "InChIKey", # InChIKey列名 (默认值)
name_col = "Name" # 用于进度显示的名称列 (默认值)
)关键参数说明: - name_col, cas_col, inchikey_col:
指定包含标识符的列 (默认值: “Name”, “CAS”, “InChIKey”) - verbose: 设为
FALSE 可抑制进度信息 (默认: TRUE) - cas, flavornet, synonyms,
uses: 启用特定元数据类型的提取 (全部默认: FALSE) -
函数自动处理网络错误并提供详细的进度信息
半定量工作流程分两步：select_std
根据分子描述符相似性为每个检出化合物分配最佳匹配的标准品，calculate_con
将峰面积转换为浓度。两种校准模式均可用，模式根据数据列自动检测：
| 模式 | 适用场景 | std_md 必需列 | 浓度公式 |
|---|---|---|---|
| 校准曲线 | 每个标准品有完整的多水平校准曲线 | Slope, Intercept（+ 可选 R2, Low_con, Low_area） |
(A / A_D8 − Intercept) / Slope |
| 单点校准 | 混标在单一已知浓度下测量 | STD_con, STD_area |
(A / A_D8 × STD_con) / STD_area |
校准曲线模式下负浓度时使用单点回退：(A / A_D8 × Low_con) / Low_area。
校准曲线模式 — 标准品数据必须包含
Slope、Intercept（必需），以及可选的 R2、Low_con、Low_area：
std_md <- reference_standards # Name, Slope, Intercept, R2, Low_con, Low_area, 描述符...单点校准模式 — 标准品数据必须包含 STD_con（已知浓度，如 10
µg/mL）和 STD_area（测得的峰面积）：
std_md <- allStd_md[allStd_md$Name %in% std$Name, ] |>
mutate(STD_area = std$STD[match(Name, std$Name)], # 注射的平均峰面积
STD_con = 10) |> # 混标浓度 (µg/mL)
relocate(STD_con, STD_area, .after = SMILES)根据分子描述符相似性为每个化合物分配最佳匹配的标准品。模式和响应列从
std_md 列自动检测；设置 std_res_col = FALSE 跳过特征选择。
quan_res <- select_std(
std_md = std_md,
std_res_col = "Slope", # 单点模式用 "STD_area"；省略 = 自动检测
std_md1_col = 8, # std_md 中分子描述符的起始列
data_md = target_compounds,
data_md1_col = 5, # data_md 中分子描述符的起始列
top_npct_md = 30 # 随机森林选择的描述符百分比
)sample_data <- left_join(sample_data,
quan_res[, c("Name", "Standard", "Similarity",
"Slope", "Intercept", "R2", "Low_con", "Low_area", # 校准曲线模式
"STD_con", "STD_area")], # 单点校准模式
by = "Name"
)将峰面积转换为浓度。模式根据存在的列自动检测。
concentrations <- calculate_con(
df = sample_data,
sam_weight = sample_weights, # NULL = 仅仪器浓度；见下方格式
start_col = 12 # 峰面积列的起始索引
)按样本名称汇总重复进样的均值±标准差。
final_results <- organize_con(
df = concentrations,
sam_code = sample_codes, # 见下方格式
start_col = 12,
digits = 3,
na2zero = TRUE,
bind_mean_sd = TRUE
)sam_weight — 用于计算单位样本重量浓度（仅需仪器浓度时省略）：
| 列名 | 说明 |
|---|---|
Name |
样本代码，必须匹配峰面积数据的列名 |
Weight |
样本重量（如 g） |
Volumn |
最终提取液体积（如 mL） |
换算：conc_in_sample = conc × Volumn / Weight（如 ng/g）。
sample_weights <- data.frame(
Name = c("S1", "S2", "S3"),
Weight = c(0.50, 0.52, 0.48),
Volumn = c(1.0, 1.0, 1.0)
)sam_code — 为 keep_area() 和 organize_con()
将原始代码映射为显示名称：
| 列名 | 说明 |
|---|---|
Code |
列名或 Comment 列中出现的代码 |
Name |
显示名称；共享相同名称的重复进样分组汇总 |
sample_codes <- data.frame(
Code = c("S1a", "S1b", "S2a", "S2b", "S3a"),
Name = c("S1", "S1", "S2", "S2", "S3")
)这两个函数互斥——不能顺序使用。
| 函数 | 公式 | 输出 | 适用场景 |
|---|---|---|---|
normalize_area() |
area / D8 |
归一化峰面积 | QC 监控、数据探索 |
calculate_con() |
见上方表格 | 浓度 | 校准曲线定量 |
顺序使用会导致 D8 归一化被应用两次，结果错误。根据目标二选一。
本项目采用MIT许可证 - 详情请参阅 LICENSE 文件。
苏启枝 (Qizhi Su) - 包开发者 - 📧 邮箱: sukissqz@gmail.com - 🆔 ORCID: 0000-0002-8124-997X - 🐙 GitHub: @QizhiSu
