Nanomaterial read across predictions with nano-lazar

Christoph Helma, Micha Rautenberg, Denis Gebele

in silico toxicology gmbh, Basel, Switzerland

eNanoMapper

`lazar` read across framework

A reproducible version of the read across procedure commonly used in toxicological risk assessment (based on the k-nearest-neighbor algorithm)

Search in a database for similar nanoparticles (neighbors)
Build a local QSAR model with these neighbors
Use this model to predict the activity of the query substance

lazar was originally designed for small molecules with a defined chemical structure. The nanoparticle extension was developed and validated within the eNanoMapper project.

Similarity calculation

Requirements

Descriptors (features) for the query substance and the neighbor candidate

Observation

A large number of irrelevant features can lead do meaningless similarity estimates

Relevant features

Features that correlate significantly with toxicity (Pearson correlation p-value < 0.05)

Weighted cosine similarity

Scaled and centered relevant feature vectors
Feature contributions weighted by Pearson correlation coefficient
Similarity threshold: sim > 0.5

Local regression algorithms

Weighted average
Weighted partial least squares regression
Weighted random forests

Partial least squares and random forest models use the caret R package with default settings

Prediction intervals: 1.96*RMSE of carets bootstrapped model predictions

If PLS/RF modelling or prediction fails, lazar resorts to using the weighted average method.

Validation

3 repeated 10-fold crossvalidations with independent training/test set splits
No fixed random seed for training/test set splits, to avoid overfitting and to demonstrate the variability of validation results due to random training/test splits.
Separate feature selection for each training dataset to avoid overfitting

Data requirements

At least 100 examples per toxicity endpoint for statistically meaningful validation results
At least non-empty intersection of descriptors for calculation of similarities

Net cell association endpoint of the Protein corona dataset (121 gold and silver particles)

10-fold crossvalidations

Descriptors	Algorithm	r²	RMSE
Physchem	WA	`0.42, 0.46, 0.48`	`2.02, 1.94, 1.92`
Physchem	PLS	`0.53, 0.54, 0.49`	`1.83, 1.8, 1.9`
Physchem	RF	`0.53, 0.52, 0.54`	`1.82, 1.84, 1.79`
Proteomics	WA	`0.66, 0.63, 0.63 *`	`1.58, 1.62, 1.66 *`
Proteomics	PLS	`0.59, 0.66, 0.63 *`	`1.74, 1.56, 1.65 *`
Proteomics	RF	`0.66, 0.65, 0.63 *`	`1.56, 1.59, 1.64 *`
All	WA	`0.73, 0.66, 0.66 *`	`1.41, 1.57, 1.58 *`
All	PLS	`0.67, 0.64, 0.69 *`	`1.53, 1.63, 1.5 *`
All	RF	`0.69, 0.69, 0.7 **`	`1.51, 1.5, 1.46 **`

Gold and silver particles included!

Correlation plot

Correlation of log2 transformed net cell association measurements with random forest predictions using physchem properties and protein corona data.

Links

Nano-lazar GUI: https://nano-lazar.in-silico.ch
Lazar (source code): https://github.com/opentox/lazar
Presentation (source code): https://github.com/opentox/nano-lazar-paper
Docker image: https://hub.docker.com/r/insilicotox/nano-lazar-paper/
Nano-lazar development version: https://nano-lazar-dev.in-silico.ch/predict

Exercises

Try the nano-lazar versions at

Old (stable) version (physchem only): https://nano-lazar.in-silico.ch
Next release: https://nano-lazar-dev.in-silico.ch/predict

Questions

Do you think that nanoparticle predictions based on physchem parameters are a practical approach
Do you think that nanoparticle predictions based on proteomics measurements are a practical approach
What would you expect from a nanoparticle read-across application
- User input
- Prediction output
Comments, bug reports and feature suggestions