Q&A 18 How do you train and evaluate a Random Forest classifier on microbiome data?

18.1 Explanation

The Random Forest algorithm is a popular and robust model for microbiome classification tasks due to its: - Built-in feature importance - Resistance to overfitting - Non-linear modeling capability

This Q&A demonstrates how to: - Train a Random Forest classifier - Evaluate it using accuracy and confusion matrix - Inspect important OTUs

18.2 Python Code

import pandas as pd
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score, confusion_matrix
from sklearn.model_selection import train_test_split

# Load and prepare data
otu_df = pd.read_csv("data/otu_table_filtered.tsv", sep="\t", index_col=0).T
meta_df = pd.read_csv("data/sample_metadata.tsv", sep="\t")
data = pd.merge(otu_df, meta_df, left_index=True, right_on="sample_id")
X = data[otu_df.columns]
y = data["group"].map({"Control": 0, "Treatment": 1})
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)

# Train Random Forest
clf = RandomForestClassifier(n_estimators=100, random_state=42)
clf.fit(X_train, y_train)

# Predict and evaluate
y_pred = clf.predict(X_test)
acc = accuracy_score(y_test, y_pred)
cm = confusion_matrix(y_test, y_pred)

print(f"Accuracy: {acc:.2f}")
print("Confusion Matrix:")
print(cm)

# Top 5 important OTUs
feat_imp = pd.Series(clf.feature_importances_, index=X.columns)
print("Top OTUs:
", feat_imp.sort_values(ascending=False).head())

18.3 R Code (caret)

library(tidyverse)
library(caret)

otu_df <- read.delim("data/otu_table_filtered.tsv", row.names = 1)
meta_df <- read.delim("data/sample_metadata.tsv")
otu_df <- otu_df[, meta_df$sample_id]
otu_df <- t(otu_df)
data <- cbind(as.data.frame(otu_df), group = meta_df$group)

# Encode group and split
data$group <- as.factor(data$group)
set.seed(42)
trainIndex <- createDataPartition(data$group, p = .7, list = FALSE)
train <- data[trainIndex, ]
test  <- data[-trainIndex, ]

# Train Random Forest
rf_model <- train(group ~ ., data = train, method = "rf", trControl = trainControl(method = "cv", number = 5))

# Predict and evaluate
pred <- predict(rf_model, newdata = test)
confusionMatrix(pred, test$group)

Confusion Matrix and Statistics

           Reference
Prediction  Control Treatment
  Control         0         1
  Treatment       1         0
                                     
               Accuracy : 0          
                 95% CI : (0, 0.8419)
    No Information Rate : 0.5        
    P-Value [Acc > NIR] : 1          
                                     
                  Kappa : -1         
                                     
 Mcnemar's Test P-Value : 1          
                                     
            Sensitivity : 0.0        
            Specificity : 0.0        
         Pos Pred Value : 0.0        
         Neg Pred Value : 0.0        
             Prevalence : 0.5        
         Detection Rate : 0.0        
   Detection Prevalence : 0.5        
      Balanced Accuracy : 0.0        
                                     
       'Positive' Class : Control