Software
Listed below is only a very small selection of software projects I am involved in. To see a complete list, please take a look at my GitHub account at https://github.com/rasbt and the [Code] links provided on the Publications pages.
Machine Learning & Deep Learning
A flexible method for ordinal regression with deep neural networks overcoming the restrictions of CORAL.
- Xintong Shi, Wenzhi Cao, and Sebastian Raschka (2021).
Deep Neural Networks for Rank-Consistent Ordinal Regression Based On Conditional Probabilities.
Arxiv preprint; https://arxiv.org/abs/2111.08851
A new method for ordinal regression with deep neural networks, addressing the rank inconsistency issue of other ordinal regression frameworks.
- Wenzhi Cao, Vahid Mirjalili, and Sebastian Raschka (2020)
Rank Consistent Ordinal Regression for Neural Networks with Application to Age Estimation
Pattern Recognition Letters. 140, 325-331
MLxtend (machine learning extensions) is a Python library of useful tools for the day-to-day data science tasks.
- Raschka, Sebastian (2018) MLxtend: Providing machine learning and data science utilities and extensions to Python’s scientific computing stack. J Open Source Softw 3(24).
Various deep learning models implemented in PyTorch and TensorFlow.
Semi-Adversarial Neural Networks Implementation
Implementation of the SAN architecture and model for imparting gender privacy to face images.
- Mirjalili, Vahid, Sebastian Raschka, Anoop Namboodiri, and Arun Ross (2018) Semi-adversarial networks: Convolutional autoencoders for imparting privacy to face images. Proc. of 11th IAPR International Conference on Biometrics (ICB 2018), Gold Coast, Australia. (Best Paper Award)
Python Utilities
Mputil is a library that provides functions for memory-efficient multi-processing, based Python’s multiprocessing standard library.
An IPython magic extension for printing date and time stamps, version numbers, and hardware information to aid reproducible research.
Utility functions for parsing BibTeX files and creating citation reference lists.
Computational Biology/Bioinformatics
Biopandas is a Python library for working with molecular structures in pandas DataFrames.
- Raschka, Sebastian (2017) BioPandas: Working with molecular structures in pandas DataFrames. J Open Source Softw 2(14).
ScreenLamp is a Python toolkit that enables the hypothesis-driven, ligand-based screening of large molecule libraries containing millions of compounds as well as the generation of molecular fingerprints for machine learning and data mining applications.
- Raschka, Sebastian, Anne M. Scott, Nan Liu, Santosh Gunturu, Mar Huertas, Weiming Li, and Leslie A. Kuhn (2018) “Enabling the hypothesis-driven prioritization of ligand candidates in big databases: Screenlamp and its application to GPCR inhibitor discovery for invasive species control”. Journal of Computer-Aided Molecular Design.
A novel approach to pose selection in protein-ligand docking based on graph theory. SiteInterlock is a Python package for selecting near-native protein-ligand docking poses based upon the hypothesis that interfacial rigidification of both the protein and ligand prove to be important characteristics of the native binding mode and are sensitive to the spatial coupling of interactions and bond-rotational degrees of freedom in the interface.
- Raschka, Sebastian, Joseph Bemister‐Buffington, and Leslie A. Kuhn. “Detecting the native ligand orientation by interfacial rigidity: SiteInterlock.” Proteins: Structure, Function, and Bioinformatics 84.12 (2016): 1888-1901.
Protein Recognition Index (PRI)
The Protein Recognition Index (PRI) measures the similarity between H-bonding features in a given complex (predicted or designed) and the characteristic H-bond trends from crystallographic complexes based on hydrogen-bond interactions identified by Hbind (software accompanying the paper for rigorously defining intermolecular H-bonds by donor/acceptor chemistry and geometric constraints).
- Raschka, Sebastian, Alex Wolf, Joseph Bemister‐Buffington, and Leslie A. Kuhn (2018) “Protein-ligand interfaces are polarized: discovery of a strong trend for intermolecular hydrogen bonds to favor donors on the protein side with implications for predicting and designing ligand complexes” Journal of Computer-Aided Molecular Design. Journal of Computer-Aided Molecular Design
Hbind – Identifying hydrogen bonds by donor/acceptor chemistry and geometric constraints
Software to rigorously define intermolecular H-bonds by donor/acceptor chemistry and geometric constraints, which was developed, used, and described in detail in
- Raschka, Sebastian, Alex Wolf, Joseph Bemister‐Buffington, and Leslie A. Kuhn (2018) “Protein-ligand interfaces are polarized: discovery of a strong trend for intermolecular hydrogen bonds to favor donors on the protein side with implications for predicting and designing ligand complexes” Journal of Computer-Aided Molecular Design. Journal of Computer-Aided Molecular Design