“IDIBAPS is the perfect bridge between clinical care, biology and computational research”

I was trained as a medical doctor and did my undergrad in Kiel, Heidelberg (Germany), with some studies abroad in Spain, the UK, the US and Australia. I did my doctoral thesis in basic immunology research, but when I started my residency, I switched to computational oncology, and that's what I essentially stuck to.

During a research project with Padmanee Sharma at the University of Texas MD Anderson Cancer Center, I started my transition from immunology to computational research. I did my postdoc with Dana Pe'er, a very renowned computational biologist who works at Sloan Kettering Institute in New York, and then I tried to combine these parts: the work as a medical oncologist, with the immunological wet lab background, and the computational part. 

The fascinating thing is that you can have very broad impact and you can move things fast, it’s very scalable. When you're working as a physician, you see one patient at a time... there's only so many patients you can see in a day. So, this scales very poorly, it scales linearly. Spend twice as long in the hospital and you can see twice as many patients. And the same goes for the wet lab. But in computational research, it's different. It's more disconnected. It's not how much time you spend; it's whether you addressed the right problem. Did you ask the right question? Did you choose the right implementation? Suddenly, this becomes hugely disproportional to your invested time. You can make a fundamental difference and change many areas of research.

Yes, I tend to. But all my projects now include a clinical component, an immunological component, and a computational component. Of course, over the years I have been putting more focus on the computational aspects, because this is where I think we can make the biggest difference.

IDIBAPS has the right size, and the connection between clinical, biological and computational research is extremely agile. You cannot provide that in a larger institution. Moreover, there is an incredible team spirit where everyone is working on a common goal and there is an unparalleled degree of collaboration between the labs. It works very well. 

So, you can always have more infrastructure on the technical and biological side, and that's certainly something where IDIBAPS cannot compete, but other strengths make up for that.

My position was an international call from the AECC program, which is funding the excellence in cancer immunotherapy area here at IDIBAPS. I am very happy my now fellow researchers chose me for this position. The whole immunotherapy team is stellar, and I hope to further reinforce this team. What we want to do is to extend the computational aspects of this program by leveraging modern machine learning and artificial intelligence to build the next generation of immunotherapies and immunodiagnostics for cancer patients.

We have just submitted our first preprint, and we are on a good track. We will submit our first report to our funders, and there are many grants pending. My team is shaping up; we have already hired three researchers and have established great collaborations on cancer immunotherapy (e.g. with Francesc Balaguer and Aleix Prat). Moreover, we are actively working with our research compute platform to enable all groups at IDIBAPS to generate the most accurate insights from their data. The rise of agentic artificial intelligence is a great opportunity to democratize data analysis. So, it's picking up speed, and I think it's very exciting.

If you're implementing something new, there are always challenges; there are always concerns and administrative barriers, because these things hadn't been anticipated before. But that's what everybody's facing. The whole field of AI has finally arrived in the public mind, it's just accelerating, and our current frameworks are a little bit rigid for that.

At my previous institution, I already had quite a lot of doctoral students, so I scaled down a little bit here. But the nice part is that here I can have more focus on research, whereas before it was more a clinical focus. I also changed my research focus: Apart from immunotherapy, we’ve moved into cancer prevention to early detect and ideally intercept melanoma and colorectal cancer.

I do speak Spanish, but I don't speak Catalan yet. I encourage my lab to speak English because it's more inclusive for our international hires, and it's also the language of science now. It’s important that we know how to communicate well, and not in a written way, but in a personal way: how do we speak in conferences, how do we bring the message across, it’s important.

There are three broad aspects. First, accelerating clinical workflows. Nowadays you can query, for example, clinical study protocols much more easily if you have implemented the right tools, and you can also automate some part of the writing and the documentation requirements. Also, we are now automating some areas of diagnostic medicine, like radiology or pathology. And now we are also building new diagnostic which allow us to guide our patients’ treatment. Machine learning and AI is helping us to make such decisions from hundreds or thousands of markers.  This allows us to ask questions like: Can we understand better how patients respond to therapy? if some people will develop cancer? Which patients do we have to monitor more closely? Do we have to give them some preventive medicines? Do we have to treat them more intensively, or less intensively? Historically, we have always had one marker: up is good and down is bad, or the other way around. 

Finally, there is therapeutics and drug design. We can now build computer models of diseases and ask: When do you give the immunotherapy? How do you combine it with other drugs?  Because the number of drug combinations is just too big to test in clinical trials these models are letting us prioritize targets. And then maybe there are additional targets we haven't thought of which we discover in these analyses.

We are a research institution; the main goal is to provide a first proof of principle that it is possible. And we have several data sets from clinical trials where we see very promising results where we think we have found a signal which could give rise to something like that. But this is, of course, very early in the process. 

In my view, this is really what a research institution is supposed to provide. And once you have this kind of proof of principle, then you can do the hard part afterwards, where you usually need industry support, where you need the pharmaceutical industry to really scale this project up. And this is also very costly and very time extensive.

We want to define which immune responses predict cancer immunotherapy outcomes across patients. We want to understand the different types of immune responses people can develop, and how these are linked to benefit or no benefit from therapy. Moreover, we will use these responses to detect cancers early. 

Lastly, we want to build the basis for the next generation of cancer vaccines. We are doing it with Lynch syndrome, and together with Francesc Balaguer, because right now, we're essentially giving these vaccines, but we are not really understanding how they exactly shape our immune cells. IDIBAPS has an outstanding trial and infrastructure, and we will make sure that the next generation of vaccines will even be better based on our results. 

I mainly came for the clinical collaborations and the strength of the translational research. But I heavily use the compute resources, which are currently provided by the Barcelona Supercomputing Center (BSC). While most of our compute is outsource, we are actively engaged with the team to build local infrastructure at IDIBAPS because I think this is something which is fundamentally important for the institution, to have some sovereignty and independence over compute resources.

Excellence, collaboration, and agility.