Better proteins. Faster.
We accelerate protein research with state of the art
artificial intelligence
Better proteins. Faster.
We accelerate protein research with state of the art
artificial intelligence
The fundamentals of protein structure and dynamics complicate drug design and industrial applications of proteins. Thus, advancing basic knowledge of proteins is pivotal for disease research and green chemistry solutions.
Fundamentals are obscured by the immense growth of public protein data. As a result, there is only stagnant growth of our knowledge, prediction models, and methods to advance protein biotech. This is why making sense of the data is paramount.
Developing models that predict or explain features of proteins are extremely complicated, yet required for smarter engineering and production. Big data is available to train models, but an effective digest of the data is lacking.
A key solution for bringing complex prediction models to protein biotech is through machine learning and AI. This fuels our mission to accelerate protein research, which is why we built the world's first protein database dSPP tailored for Deep Learning and AI applications with full Keras™ and Tensorflow™ integration.
The fundamentals of protein structure and dynamics complicate drug design and industrial applications of proteins. Thus, advancing basic knowledge of proteins is pivotal for disease research and green chemistry solutions.
Fundamentals are obscured by the immense growth of public protein data. As a result, there is only stagnant growth of our knowledge, prediction models, and methods to advance protein biotech. This is why making sense of the data is paramount.
Developing models that predict or explain features of proteins are extremely complicated, yet required for smarter engineering and production. Big data is available to train models, but an effective digest of the data is lacking.
A key solution for bringing complex prediction models to protein biotech is through machine learning and AI. This fuels our mission to accelerate protein research, which is why we built the world's first protein database dSPP tailored for Deep Learning and AI applications with full Keras™ and Tensorflow™ integration.
Proteins are complex and heterogeneous biomolecules. They owe their complicated nature to a combination of structure and dynamics. We are working on the prediction of structural features and intrinsic dynamics from their sequences and known structural models.
Intrinsic protein disorder is difficult to model, yet necessary for advancing protein structure, function and dynamics prediction. Our algorithms can quantify protein disorder at the residue level with unprecedented level of accuracy thanks to inclusion of dSPP™ data.
Thermal stability influences protein production on the laboratory and industrial scale. Our research into intrinsic protein dynamics and protein sequence-structure relationships yields predictive models, which will help to identify production bottlenecks and gain high protein yields.
Protein electrostatics has profound consequences for prediction of structure, dynamics and stability. Our propriety, hybrid machine learning algorithms can predict electrostatic effects of single mutations on protein stability, isoelectric point (pI) and optimal pH. Therefore, production and purification protocols can be fine tuned for individual protein mutants.
Laboratory evolution of proteins made a seminal impact for protein engineers in the last two decades. Now, we must harness machine learning to advance protein engineering to the next level. Through streamlined integration of our prediction tools, we are able to perform in silico evolution for smarter engineering.
Our dSPP database can be effortlessly setup and integrated with Keras™ and Tensorflow™ using Python pip manager.
pip install dspp-keras
To explore the full potential of dSPP, train example models from our Github repository.
Learn moreOur dSPP database can be effortlessly setup and integrated with Keras™ and Tensorflow™ using Python pip manager.
pip install dspp-keras
To explore the full potential of dSPP, train example models from our Github repository.
We provide private research and consultancy for academic and industrial clients. Our research process includes: stringent statistical analyses of input data with thorough assessment of errors, state of the art numerical modelling supported by Machine Learning and Artificial Intelligence techniques, and training and development of powerful and intuitive tools that are complementary to existing software stack.
Please read dSPP™ research paper to see our workflow.
Read paper
We provide private research and consultancy for academic and industrial clients. Our research process includes: stringent statistical analyses of input data with thorough assessment of errors, state of the art numerical modelling supported by Machine Learning and Artificial Intelligence techniques, and training and development of powerful and intuitive tools that are complementary to existing software stack.
Please read dSPP™ research paper to see our workflow.
Kamil Tamiola is the architect of Peptone, leveraging a profound understanding of intricate workings of machine learning and high performance computing techniques.
Matthew Heberling brings an invaluable blend of problem solving skills and professional experience to Peptone with his business and biotechnology background.
Jan Domanski provides critical scientific insight and expertiese born out of academic research on molecular biophysics and computational protein dynamics.
Kamil was born to a family of medical doctors. Inspired by the 1st edition of Feynman’s lectures on physics, Kamil decided to pursue a purely scientific path, a deed considered as a rebellious act by his parents, given the medical traditions in the family.
The passion for applied physics and computing were instrumental in getting the first peer-reviewed paper accepted, "Seeing and Recognising Objects as Physical Process - Practical and Theoretical Use of Artificial Neural Networks" at the age of 18. The aforementioned article, recognised by the Polish Academy of Sciences, won Kamil an exam-free admission to University of Wroclaw, where he conduced personalised research in the area of computational biophysics with the specialisation in numerical modelling of proteins. Shortly after followed the undergraduate research at Cambridge University with the prestigious Bill Gates foundation scholarship, and the admittance to an elite master degree programme, “Top Master in Bimolecular Sciences”, at University of Groningen, the Netherlands.
Working under the watchful eye of researchers in the Biotechnology Laboratory of Groningen Bimolecular Sciences Institute (GBB), Kamil developed his first industrial-scale numerical models, which have direct implications in the production of an antibiotic: penicillin. His curiosity and passion for high performance computing led him to the laboratory of Molecular Dynamics and Nuclear Magnetic Resonance Spectroscopy (NMR) where he begun his doctoral studies in computational biophysics. Tutored by Prof. Frans A.A. Mulder, Kamil developed highly cited tools and relational database used to characterise a perplexing class of intrinsically disordered proteins (IDPs), directly implicated in neurodegenerative disorders, including “mad cow” disease, and Alzheimer’s.
Matt was born 1978 in small-town, U.S.A. (Ohio). He learned early-on about the power of vision in sports (cross country state), which propelled his career change from tax accountant to protein engineer. His "Ah-ha" moments that sparked this change: reflections on a childhood filled with curiosity and Legos and wondering how scientists study molecules they can't see.
Matt began research with Prof. Tom Magliery (Ohio State U.) after being awarded a scholarship to investigate tumor suppressor protein, p53. Aside from the intellectual growth, Prof. Magliery inspired Matt with effective scientific communication. Intrigued by the intricacies of protein structure-function, his next projects with Prof. Andreas Plueckthun (U. of Zurich) examined innovative approaches for protein structure determination using robust binding proteins. This experience groomed his 'think outside of the box' mentality.
His developed curiosity about proteins that catalyze reactions (enzymes) landed him a doctoral research position with Prof. Dick Janssen (Groningen U.). There, Matt expanded his knowledge into enzyme discovery and engineering, bioinformatics, metagenomics, and microbiology. He focused on the discovery of transaminases and aminomutases for the conversion of unnatural amino acids that led to 6 (co)-authored publications.
Jan obtained his MBioch from Oxford, then started his PhD with NIH/Wellcome Trust funded scholarship working on membrane protein folding. In his spare time, Jan enjoys reading, cycling and spending time with friends.
