From Failed Experiments to Predictive GPCR Models
- Dr. GPCR Podcast
- 3 hours ago
- 5 min read
From failed assays to breakthroughs in GPCR modeling, Dr. Jens Carlsson’s path into science was anything but straightforward. When he first began working in a lab, success seemed elusive: experiments often failed, and bench work felt unnatural. At one point, he even questioned whether research was the right career for him.
The turning point came from an unexpected source: a letter of recommendation.
A professor highlighted Carlsson’s talent in molecular modeling, a skill he hadn’t yet recognized as central to his future.
That recognition shifted everything.
Today, Carlsson is a Professor of Computational Biochemistry at Uppsala University and one of the most respected voices in GPCR modeling, where his group uses structure-based techniques not just to explain experimental results but to predict them.
Finding Science Through Serendipity
Carlsson didn’t grow up with a vision of becoming a scientist. Raised in a small town in southern Sweden, he had never met anyone with a PhD and had little exposure to research.
When he moved to Uppsala in the late 1990s to study engineering, it was biotechnology—then on the rise—that caught his interest. Inspired by news stories about breakthroughs like the cloning of Dolly the sheep and genetically modified foods, he saw life sciences as a field full of promise.
Still, early research internships were rocky. A summer spent purifying proteins highlighted his discomfort at the bench. While others refined lab techniques, he found himself gravitating toward structural models in his spare time. He was naturally drawn to analyzing solved protein structures—an activity that, unbeknownst to him, was laying the groundwork for a future in computational modeling.
Discovering Modeling (and GPCRs) by Accident
That interest led him to pursue a thesis at Scripps Research in San Diego, where he focused on how proteins respond to pH changes using molecular simulations. It was the first time he found himself immersed in a computational environment, and he realized how much he enjoyed the intellectual energy of modeling communities.
When he returned to Sweden for his PhD, he focused on small molecule design. At the time, there were no solved structures of GPCRs readily available, so his work remained disconnected from that target class. It wasn’t until his postdoctoral years at UCSF—under Brian Shoichet—that GPCRs entered his scientific view. His introduction came through a practical suggestion: explore a new protein family that was gaining traction in the structural biology world.
At that point, very few GPCR crystal structures had been determined. This scarcity made the field both exciting and high-risk. With limited structural data but a growing pharmacological interest, GPCRs presented the perfect challenge for someone who wanted to build predictive models from scratch.
Predictive GPCR-Ligand Modeling
Carlsson's work quickly shifted from curiosity to impact. One of his early projects involved the A2A adenosine receptor, a GPCR with known relevance in diseases like Parkinson’s.
Using virtual screening, he was able to identify novel ligands that aligned with experimental findings.
This success was a revelation—it was possible not only to interpret experimental data but to forecast it through modeling. This realization sparked a new guiding principle: computational tools should aim to predict experimental outcomes.
For Carlsson, this marked a shift in how his lab approached GPCR research. Rather than focusing solely on explaining receptor behavior post hoc, his group began developing workflows and strategies that could drive experimental design forward.
Bridging Computation and Collaboration
While Carlsson began his scientific career with a deep skepticism about collaborations—particularly with experimentalists—his experiences in the GPCR field forced a reevaluation. He came to see that effective GPCR research requires true interdisciplinary integration.
Collaborating with chemists, biologists, and pharmacologists not only made his predictions more useful, but also shaped the kinds of questions his lab could ask.
Today, his group includes around ten computational chemists and one in-house medicinal chemist. The division of labor reflects a pragmatic approach to problem-solving.
While early stages of a project often rely on virtual screening from commercial compound libraries, the work inevitably reaches a point where novel, non-commercial ligands are needed. This is when the synthetic expertise of the chemist becomes essential.
Pharmacology, on the other hand, is often outsourced through collaborations with expert labs who specialize in particular receptors.
Carlsson notes that it’s easier to find pharmacologists to run assays than it is to persuade chemists to synthesize new compounds based on computational predictions—a sentiment that highlights the skepticism that still exists around modeling in some corners of drug discovery.
Embracing Complexity and Failure
Carlsson’s approach to modeling is rooted in scientific humility. He emphasizes that not every question can be answered computationally—and that saying “we don’t know” is a valid, and often necessary, scientific position.
When asked whether a compound is twice as potent as another, he’s quick to point out the limitations in both experimental and computational resolution. This perspective influences how his lab trains students. It's not enough to run simulations or generate models.
Trainees must learn how to interpret assay data, understand pharmacological context, and communicate across disciplines.
Many of his former students now work in industry, where their ability to bridge the computational-experimental divide makes them highly valuable.
Career Lessons for Young Scientists
Carlsson’s journey holds several lessons for early-career researchers.
First and foremost: follow the questions that keep you up at night.
He believes genuine interest—not trends or external validation—is what sustains long-term scientific productivity. The daily failures and rejections of research require an intrinsic motivation that goes beyond job titles or metrics. He also underscores the importance of mentorship—not just for guidance, but for perspective.
Good mentors help shape thinking without prescribing decisions.
For Carlsson, influential mentors helped him find confidence in his own scientific voice while remaining open to other perspectives.
Finally, he advises young scientists to celebrate small wins. In a field where major publications and grant awards can be rare, finding satisfaction in an optimized curve, a new insight, or a well-modeled structure is what keeps momentum going.
AI and the Future of GPCR Modeling
Carlsson is excited by the potential of tools like AlphaFold and the continued evolution of AI in structure prediction. His lab has already begun using AlphaFold models to identify ligands for targets that lack experimental structures.
While these tools aren’t perfect—and sometimes fail in unexpected ways—they represent a shift in what’s possible.
Still, he emphasizes that data limitations remain a major challenge, especially when it comes to small molecule prediction. Unlike protein sequences, small molecule data are often fragmented, inconsistent, or unavailable, making model training more difficult.
Until better datasets emerge, predictive modeling will continue to rely on creative integration of computational and experimental insights.
Modeling a Career on Your Own Terms
Carlsson’s career shows that failures can evolve into strengths and that computational insights can transform how we approach GPCRs.
As predictive modeling matures, its role will continue to expand, guiding ligand discovery, informing pharmacology, and accelerating translation into the clinic.
For early-career researchers, the takeaway is direct: GPCR drug discovery will increasingly depend on those who can unite modeling with experimentation, turning predictions into real therapeutic breakthroughs.
Hear the full conversation with Jens Carlsson
💡 Stay curious. Stay connected.
Looking for more insights like this? The Dr. GPCR Premium Ecosystem gives you exclusive access to thought leaders, technical resources, and deep-dive content you won’t find anywhere else.
🔬 Fuel your science. 🤝 Grow your network. 🚀 Lead the future.
Comments