Comparison Workflow Notebooks

Inspect the notebooks associated with the imputation workflow.

Split creation and data handling

• Transfer data for NAGuideR format
• Experiment 03 - Data
  • Arguments
  • Raw data
  • Machine metadata
  • Select a subset of samples if specified (reduce the number of samples)
  • First Step: Select features by prevalence
  • Second step - Sample selection
  • Sample Medians and percentiles
  • Feature frequency in data
  • Split: Train, validation and test data
  • plot distribution of splits
  • Save parameters
  • Saved Figures

PIMMS models

• CF
  • Load data in long format
  • Collaborative Filtering
  • Data in wide format
  • Save predictions
  • Config
• VAE
  • Load data in long format
  • Initialize Comparison
  • Data in wide format
  • Variational Autoencoder
  • Comparisons
  • Save predictions
  • Config
• Median
  • Load data in long format
  • Initialize Comparison
  • Data in wide format
  • Comparisons
  • Save predictions
  • Config
• DAE
  • Load data in long format
  • Data in wide format
  • Denoising Autoencoder
  • Comparisons
  • Save predictions
  • Config
• KNN5
  • Load data in long format
  • Initialize Comparison
  • Data in wide format
  • Train
  • Comparisons
  • Save predictions
  • Config
• KNN
  • Load data in long format
  • Initialize Comparison
  • Data in wide format
  • Train
  • Comparisons
  • Save predictions
  • Config

R models

• IMPSEQ
  • Parameters
  • Dump predictions
• ZERO
  • Parameters
  • Dump predictions
• SVDMETHOD
  • Parameters
  • Dump predictions
• MLE
  • Parameters
  • Dump predictions
• IMPSEQROB
  • Parameters
  • Dump predictions
• IMPUTE
  • Parameters
  • Dump predictions
• BPCA
  • Parameters
  • Dump predictions
• PI
  • Parameters
  • Dump predictions
• MINDET
  • Parameters
  • Dump predictions
• QRILC
  • Parameters
  • Dump predictions
• MINPROB
  • Parameters
  • Dump predictions
• RF
  • Parameters
  • Dump predictions
• ROWMEDIAN
  • Parameters
  • Dump predictions
• IRM
  • Parameters
  • Dump predictions
• transfer NAGuideR pred
  • Test Datasplit
• COLMEDIAN
  • Parameters
  • Dump predictions
• LLS
  • Parameters
  • Dump predictions
• TRKNN
  • Parameters
  • Dump predictions
• MINIMUM
  • Parameters
  • Dump predictions

Imputation model comparison

• Compare models
  • data completeness across entire data
  • Model specifications
• Load predictions on validation and test data split
  • Validation data
  • test data
  • Figures dumped to disk

Differential Analysis Notebooks

Inspect the notebooks associated with the differential analysis workflow.

Differential analysis (ANCOVA)

• QRILC
  • Parameters
  • Data
  • ALD study approach using all measurements
  • Load model predictions for (real) missing data
  • Mean shift by model
  • Differential analysis
• VAE
  • Parameters
  • Data
  • ALD study approach using all measurements
  • Load model predictions for (real) missing data
  • Mean shift by model
  • Differential analysis
• DAE
  • Parameters
  • Data
  • ALD study approach using all measurements
  • Load model predictions for (real) missing data
  • Mean shift by model
  • Differential analysis
• None
  • Parameters
  • Data
  • ALD study approach using all measurements
  • Load model predictions for (real) missing data
  • Mean shift by model
  • Differential analysis
• CF
  • Parameters
  • Data
  • ALD study approach using all measurements
  • Load model predictions for (real) missing data
  • Mean shift by model
  • Differential analysis
• PI
  • Parameters
  • Data
  • ALD study approach using all measurements
  • Load model predictions for (real) missing data
  • Mean shift by model
  • Differential analysis
• TRKNN
  • Parameters
  • Data
  • ALD study approach using all measurements
  • Load model predictions for (real) missing data
  • Mean shift by model
  • Differential analysis
• RF
  • Parameters
  • Data
  • ALD study approach using all measurements
  • Load model predictions for (real) missing data
  • Mean shift by model
  • Differential analysis
• Median
  • Parameters
  • Data
  • ALD study approach using all measurements
  • Load model predictions for (real) missing data
  • Mean shift by model
  • Differential analysis

Compare ANCOVAs

• PI vs Median
  • Parameters
  • Load scores
  • Describe scores
  • Load frequencies of observed features
  • Compare shared features
  • Plot qvalues of both models with annotated decisions
  • Only features contained in model
  • DISEASES DB lookup
  • Shared features
  • only by model
  • Only by model which were significant
  • Shared which are only significant for by model
  • Only significant by RSN
  • Write to excel
  • Outputs
• PI vs VAE
  • Parameters
  • Load scores
  • Describe scores
  • Load frequencies of observed features
  • Compare shared features
  • Plot qvalues of both models with annotated decisions
  • Only features contained in model
  • DISEASES DB lookup
  • Shared features
  • only by model
  • Only by model which were significant
  • Shared which are only significant for by model
  • Only significant by RSN
  • Write to excel
  • Outputs
• PI vs CF
  • Parameters
  • Load scores
  • Describe scores
  • Load frequencies of observed features
  • Compare shared features
  • Plot qvalues of both models with annotated decisions
  • Only features contained in model
  • DISEASES DB lookup
  • Shared features
  • only by model
  • Only by model which were significant
  • Shared which are only significant for by model
  • Only significant by RSN
  • Write to excel
  • Outputs
• PI vs RF
  • Parameters
  • Load scores
  • Describe scores
  • Load frequencies of observed features
  • Compare shared features
  • Plot qvalues of both models with annotated decisions
  • Only features contained in model
  • DISEASES DB lookup
  • Shared features
  • only by model
  • Only by model which were significant
  • Shared which are only significant for by model
  • Only significant by RSN
  • Write to excel
  • Outputs
• PI vs TRKNN
  • Parameters
  • Load scores
  • Describe scores
  • Load frequencies of observed features
  • Compare shared features
  • Plot qvalues of both models with annotated decisions
  • Only features contained in model
  • DISEASES DB lookup
  • Shared features
  • only by model
  • Only by model which were significant
  • Shared which are only significant for by model
  • Only significant by RSN
  • Write to excel
  • Outputs
• PI vs DAE
  • Parameters
  • Load scores
  • Describe scores
  • Load frequencies of observed features
  • Compare shared features
  • Plot qvalues of both models with annotated decisions
  • Only features contained in model
  • DISEASES DB lookup
  • Shared features
  • only by model
  • Only by model which were significant
  • Shared which are only significant for by model
  • Only significant by RSN
  • Write to excel
  • Outputs
• PI vs QRILC
  • Parameters
  • Load scores
  • Describe scores
  • Load frequencies of observed features
  • Compare shared features
  • Plot qvalues of both models with annotated decisions
  • Only features contained in model
  • DISEASES DB lookup
  • Shared features
  • only by model
  • Only by model which were significant
  • Shared which are only significant for by model
  • Only significant by RSN
  • Write to excel
  • Outputs

Compare single differential analysis

• Compare single features
  • Parameters
  • Load scores
  • Plots for inspecting imputations (for diverging decisions)
  • Load target
  • Measurments
  • Compare with target annotation

Logistic regression models

• PI vs Median
  • Parameters
  • Load data
  • Modeling setup
  • Determine best number of parameters by cross validation procedure
  • Train, test split
  • Results
  • Train data plots
• PI vs VAE
  • Parameters
  • Load data
  • Modeling setup
  • Determine best number of parameters by cross validation procedure
  • Train, test split
  • Results
  • Train data plots
• PI vs CF
  • Parameters
  • Load data
  • Modeling setup
  • Determine best number of parameters by cross validation procedure
  • Train, test split
  • Results
  • Train data plots
• PI vs RF
  • Parameters
  • Load data
  • Modeling setup
  • Determine best number of parameters by cross validation procedure
  • Train, test split
  • Results
  • Train data plots
• PI vs TRKNN
  • Parameters
  • Load data
  • Modeling setup
  • Determine best number of parameters by cross validation procedure
  • Train, test split
  • Results
  • Train data plots
• PI vs DAE
  • Parameters
  • Load data
  • Modeling setup
  • Determine best number of parameters by cross validation procedure
  • Train, test split
  • Results
  • Train data plots
• PI vs QRILC
  • Parameters
  • Load data
  • Modeling setup
  • Determine best number of parameters by cross validation procedure
  • Train, test split
  • Results
  • Train data plots