Skip to content
4 minutes

As you may already know, AlphaFold is a much-hyped tool that applies machine learning to protein structure prediction. But the potential of machine learning for protein engineering and design doesn’t end there!

Machine learning can be used to achieve a lot of different outcomes beyond only structure prediction – such as protein docking and small molecule modelling. In our experience, applying machine learning to optimise existing processes or to re-assess existing data can also add great value for life sciences companies and provide a considerable competitive advantage. So, let’s dive into the details with a couple of case studies. Aside from modelling proteins using AlphaFold, what else can machine learning (ML) be used for in the field of life sciences?

Optimising existing pipelines and processes using machine learning

Even if development pipelines are already working out alright, the smart application of machine learning can make those processes more efficient and increase the chance of experimental success.

A great example of this is a project that BioLizard embarked on with a large biotech expert in antibody-based drug development. This project was aimed at leveraging the vast amount of in-house data to provide added value to the company, and in doing so, make the company more data-driven. On top of that, this client was also interested in applying AI to optimise ongoing processes within the company.

To accomplish this, BioLizard analysed the whole process flow from beginning to end: all the way from a specific patient or disease to the development of therapeutic antibodies. In doing so, BioLizard identified a few key opportunities for fine-tuning data management procedures in order to better capture data and subsequently enable the use of artificial intelligence.

In addition, a few bottlenecks that could be optimised by the smart use of AI were identified. One such bottleneck was widened by development of an ML algorithm to predict and identify beneficial properties of antibodies early on in the client’s production pipeline. This ML pipeline saved months of wet-lab work – thereby also saving a lot of time and money. Instead of toiling away in the lab to get to the same conclusion, applying machine learning early on in the antibody production pipeline could be used to already predict and generate protein sequences that would have the desired characteristics. This streamlined the process of deciding whether or not to continue with potential products, resulting in both improved efficiency and a higher chance of experimental success in the wet lab!

Creating custom algorithms to answer difficult questions

Have you ever read an article that describes a state-of-the-art, ML approach to answering a scientific question, and thought, “I want to do that too?”

ML_blog2_imageThat is exactly what happened in another successful client case, in which a leader in cell-based cancer therapies approached BioLizard for support. They were interested in optimising peptide sequences for specific targets, and a recent publication had inspired them to utilise AI within their product pipeline. The challenge was to improve existing in-house peptide libraries, in order to reduce time from discovery to identification of a final lead – and in doing so, reach an end-product of the highest possible quality. The task of BioLizard, therefore, was to reproduce or improve upon the published ML algorithms, in order to bring the client’s lead libraries to the next level.

While getting familiar with the type of data that the client had in-house, BioLizard saw a huge potential for using AI and ML. In the end, BioLizard developed a custom synthetic peptide library that had even more functionality than what was described in the original publication, and provided sequences with predicted properties superseding those in the client’s original library.

In the words of the client:

BioLizard has developed machine learning models to create or optimise synthetic libraries that perform even better than in vitro libraries. We’ve also applied machine learning to these libraries, which usually consist of several million sequences, to select only the most promising targets for follow-up wet-lab experiments.

- Anonymous


Machine learning meets personalised medicine

Another area in which we see great promise for the application of machine learning is in personalised medicine – and specifically in getting personalised products to market as fast as possible. Machine learning has huge potential for streamlining processes to get products to patients without delay. We predict that using AI and machine learning to facilitate this revolution in personalised medicine will also make life sciences companies more truly data driven, and accelerate their overall speed in R&D – which is of course one of the biggest key competitive advantages that a life sciences company can gain.


Are you ready to apply machine learning to fast-track your R&D processes?

Reach out to us today to discuss how we can help you pinpoint where, and how, to apply machine learning and artificial intelligence to bring more value to your life sciences company.

Recommended Reading