skip to content

Department of Chemical Engineering and Biotechnology

 
""

Research led by Duygu Dikicioglu, Nishanthi Gangadharan and Jeanet Mante from our Bioscience Engineering group, headed by Professor Nigel Slater, have been awarded the ‘Paper of the Year’ for 2019 by Springer journal Bioprocess and Biosystems Engineering.

In their paper, “A heuristic approach to handling missing data in biologics manufacturing databases”, the researchers explore how to deal with gaps in bioprocess data sets, in order to use them for machine learning models that could improve efficiency in the biotech industry.

Industrial and academic research into bioprocesses – such as the production of antibodies, enzymes or proteins – generates huge datasets, as researchers seek the optimum conditions across a myriad of variables for a particular process.

There are strict regulations for research into biologic drugs, so researchers must keep comprehensive records of all their experiments to prove their approach meets quality control measures.

Leverhulme Early Career Fellow Duygu Dikicioglu saw the potential to use this vast dataset in her group’s development of mathematical modelling algorithms designed to predict the outcomes of different reactions and conditions. The models could cut down the amount of trial and error in the bioprocess research.

“In product development, they record everything and generate these datasets, but only for the sake of complying with the Security Initiative; they don't really make use of it,” says Dikicioglu. “What we are doing is analysing that data and trying to understand whether we can come up with some patterns in the data that would allow us to predict what will happen in the future. So that you can monitor how a bioprocess is working and predict whether it will give you the desired outcome or whether you need to change the conditions and how to change them.

“This paper was actually the first stage of that process, because, as the data we collected is only there for ‘checking the boxes’ purposes, it’s not like the standard perfect data sets used for modelling. Some of these data come from years and years ago, so they have changed instruments in between and the new instruments starts recording something new – a different parameter. Or someone decides that taking samples not daily but every other day would also work well, and they record that, but that person has now left the job, so it's back to being recorded daily. There are all sorts of issues with the original data that we started working on and Nisha’s work was on how to handle this.”

A flow chart showing how the work should enable continuous monitoring and improvement of bioprocesses

Jeanet Mante, a visiting undergraduate student from the department of Plant Sciences to Dikicioglu and Slater’s group, began the work, which was then picked up by Nisha Gangadharan for her Master’s research.  Together, they developed approaches to model how the gaps in the data could be filled using a combination of simple averaging strategies and more complex function analysis and Gangadharan has continued the project for her PhD research. The work was funded by AstraZeneca (then MedImmune) and has enabled Dikicioglu and Gangadharan to begin more advanced work developing the predictive models based on their now complete data sets.

“This is a very basic introductory paper,” says Dikicioglu. “We just put it out there to let people know we're working on this. Nisha has developed very elaborate methods for handling the missing information, analysing the data and constructing the general predictive models, so there will be a continuation as a much more substantial paper.

“The award was completely unexpected; I think all the authors were really happy. When we started working on it, everybody had a feeling that this was an important thing, but at the same time, not a lot of people are working on it. So we're really happy that our work has been taken well by the community.”

“I was super excited when I saw the email,” says Gangadharan. “It’s a nice way to start the year. It’s like validation from the community that there’s a need for the work you do. There are more exciting things coming in the future that we hope to get out as quickly as possible. The work that we're doing right now: I'm expecting more enthusiasm from people who are reading it!”

"I was very happily surprised by the award,” says Jeanet. “I think it is a reflection of the wonderful mentor Duygu is, and the collaboration she fostered between Nisha and myself. Nigel and Duygu gave me the opportunity to build confidence in my research aptitude, which in turn gave me the confidence to pursue a PhD. Seeing work that meant a lot to me personally gaining recognition is especially meaningful."

Jeanet Mante is now a PhD student at the University of Utah, Nisha Gangadharan is a PhD student in our department and Duygu Dikicioglu will take up a position as Associate Professor of Digital Bioprocess Engineering at University College London in April this year.

Read the full paper: “A heuristic approach to handling missing data in biologics manufacturing databases”, 2019, Bioprocess and Biosystems Engineering.

 

 

 

 

 

 

 

 

 

 

 

Duygu Dikicioglu (left image) and Nisha Gangadharan (left of right image) and Jeanet Mante (right of right image)

Latest news

Supercharging decarbonisation through intelligent technologies

17 February 2020

Integrating digital tools into the world’s energy systems could reduce carbon emissions by more than 50 per cent, a new review has found.

Intellectual property workshop with Simmons & Simmons for our MBE students

4 February 2020

Last week our Masters in Bioscience Enterprise student cohort attended a day workshop on intellectual property at Simmons & Simmons London Headquarters office.

Filling the gaps in biologics data: bioscience engineering paper wins ‘Paper of the Year’ award

30 January 2020

Research led by Duygu Dikicioglu, Nishanthi Gangadharan and Jeanet Mante from our Bioscience Engineering group, headed by Professor Nigel Slater, has been awarded the ‘Paper of the Year’ for 2019 by Springer journal Bioprocess and Biosystems Engineering