1 Sample Information

The 77 samples present in the microarray and PCR data sets used in this analyses are listed below.

The sample group stratification used for the analyses is the following. This decision was made considering the small sample sizes on the “Diagnosis” column described below where samples have unique diagnostic labels.

2 Comparison #1: (PostTx + IFTA + BKV) vs (AR + cAMR)

2.1 Correlation Analysis

2.1.1 Microarray

2.1.2 PCR #1

2.1.3 PCR #2

2.1.4 Microarray and PCR #1

2.1.5 Microarray and dCT #2

2.1.6 dCT #1 and dCT #2

2.2 Predictive Capabilities of kSORT with Support Vector Machine

2.2.1 Microarray

2.2.2 PCR #1

2.2.3 PCR #2

2.3 Recursive Feature Elimination with SVM

Recursive Feature Elimination with SVM (RFESVM) was carried out to recursively remove features in order to obtain a minimal model with the best accuracy score. The line plots represent the accuracy value for each decreasing size of the subset of genes, additionally, a confusion matrix with the selected subset is available for each comparison. Further description of the selected genes and corresponding coefficient in the resulting model are available in the tables. The results indicate that removing features from the data model improve the mapping functions learned by SVM, especially to predict rejection - this is depicted in the confusion matrices with the selected features where there has been improvement in the correct predictions for this class.

2.3.1 Microarray

2.3.2 PCR #1

2.3.3 PCR #2

2.4 Feature Importance with Boruta

Boruta, a wrapper feature selection algorithm with Random Forest as classifier, was used to perform feature importance analysis using the kSORT gene set as features and the sample groups as classes (i.e. Control, Rejection).

2.4.1 Microarray

2.4.2 PCR #1

2.4.3 PCR #2

3 Comparison #2: (PreTx + IFTA + BKV) vs (AR + cAMR)

3.1 Correlation Analysis

3.1.1 Microarray

3.1.2 PCR #1

3.1.3 PCR #2

3.1.4 Microarray and PCR #1

3.1.5 Microarray and dCT #2

3.1.6 dCT #1 and dCT #2

3.2 Predictive Capabilities of kSORT with Support Vector Machine

3.2.1 Microarray

3.2.2 PCR #1

3.2.3 PCR #2

3.3 Recursive Feature Elimination with SVM

3.3.1 Microarray

3.3.2 PCR #1

3.3.3 PCR #2

3.4 Feature Importance with Boruta

3.4.1 Microarray

3.4.2 PCR #1

3.4.3 PCR #2

4 Comparison #3: (PreTx + PostTx + IFTA + BKV) vs (AR + cAMR)

4.1 Correlation Analysis

4.1.1 Microarray

4.1.2 PCR #1

4.1.3 PCR #2

4.1.4 Microarray and PCR #1

4.1.5 Microarray and dCT #2

4.1.6 dCT #1 and dCT #2

4.2 Predictive Capabilities of kSORT with Support Vector Machine

4.2.1 Microarray

4.2.2 PCR #1

4.2.3 PCR #2

4.3 Recursive Feature Elimination with SVM

4.3.1 Microarray

4.3.2 PCR #1

4.3.3 PCR #2

4.4 Feature Importance with Boruta

4.4.1 Microarray

4.4.2 PCR #1

4.4.3 PCR #2

5 Oversampling with ADASYN

5.1 Comparison #1.1: (PostTx + IFTA + BKV) vs (AR + cAMR)

5.1.1 Predictive Capabilities of kSORT with Support Vector Machine

5.1.1.1 Microarray

5.1.1.2 PCR #1

5.1.1.3 PCR #2

5.1.2 Recursive Feature Elimination with SVM

5.1.2.1 Microarray

5.1.2.2 PCR #1

5.1.2.3 PCR #2

5.1.3 Feature Importance with Boruta

5.1.3.1 Microarray

5.1.3.2 PCR #1

5.1.3.3 PCR #2

5.2 Comparison #2.1: (PreTx + IFTA + BKV) vs (AR + cAMR)

5.2.1 Predictive Capabilities of kSORT with Support Vector Machine

5.2.1.1 Microarray

5.2.1.2 PCR #1

5.2.1.3 PCR #2

5.2.2 Recursive Feature Elimination with SVM

5.2.2.1 Microarray

5.2.2.2 PCR #1

5.2.2.3 PCR #2

5.2.3 Feature Importance with Boruta

5.2.3.1 Microarray

5.2.3.2 PCR #1

5.2.3.3 PCR #2

5.3 Comparison #3.1: (PreTx + PostTx + IFTA + BKV) vs (AR + cAMR)

5.3.1 Predictive Capabilities of kSORT with Support Vector Machine

5.3.1.1 Microarray

5.3.1.2 PCR #1

5.3.1.3 PCR #2

5.3.2 Recursive Feature Elimination with SVM

5.3.2.1 Microarray

5.3.2.2 PCR #1

5.3.2.3 PCR #2

5.3.3 Feature Importance with Boruta

5.3.3.1 Microarray

5.3.3.2 PCR #1

5.3.3.3 PCR #2