Biochemists and Biophysicists
Scrub through 162years of this role's history, from when it first emerged, through every wave of technology that reshaped it, to the cited projections for where it's heading next.
The tools that defined the work
Select an era to see how it reshaped the work.
Wet chemistry and fermentation analysis (spectroscopy, titration, colorimetry)
The founding generation of biochemists worked with the analytical tools of late 19th-century chemistry: colorimetry, wet titration, fractional precipitation, and elementary spectroscopy. Eduard Buchner's 1897 demonstration of cell-free fermentation in yeast extract was the field's foundational experimental moment, proving that enzymatic chemistry could proceed outside a living cell and therefore be studied chemically. The techniques of this era could measure the products and stoichiometry of biochemical reactions but could not reveal the molecular structures responsible for them. Equipment was simple glassware, waterbaths, and beam balances; the limiting factor was the scientist's ability to design a clean chemical assay for a biological process.
Work toolChanging equipment Enzyme crystallization, ultracentrifuge, and paper chromatography
James Sumner's 1926 crystallization of urease proved that enzymes are pure proteins and gave structural biochemistry its first toehold. The Svedberg ultracentrifuge (introduced 1924) allowed determination of protein molecular weights for the first time. Paper chromatography (Martin and Synge, 1941 Nobel 1952) transformed amino acid and small-molecule separation from a days-long effort to a single-afternoon technique. These tools let biochemists begin characterizing the molecular composition of enzymes and metabolic intermediates, building toward the complete descriptions of the citric acid cycle (Krebs, 1937) and the metabolic map. The era's hallmark was the move from measuring what reactions happened to identifying which molecules were involved.
Effect on the workThese analytical advances created demand for PhD biochemists across university medical schools and the growing pharmaceutical industry. By 1950 biochemistry departments existed at all major research universities and medical schools, and industrial employment (at Eli Lilly, Merck, Pfizer, and a dozen others) was growing rapidly around vitamins, antibiotics, and hormone production.
Work toolChanging equipment X-ray crystallography and the DNA double helix (structural biology era 1.0)
X-ray crystallography had been applied to small molecules since the Braggs' 1912-1913 work, but it became a biology tool in the 1950s and 1960s when Rosalind Franklin's fiber diffraction data helped reveal the DNA double helix (1953) and David Phillips's group solved the first protein crystal structure at atomic resolution, lysozyme, in 1965. The Protein Data Bank was founded in 1971 with seven initial structures. By 1980 it held a few hundred; by 1990 a few thousand; by 2000 over 10,000. X-ray crystallography transformed biochemistry's central question: not just what molecules are involved in a biological process, but what three-dimensional architecture makes them work. The era created the discipline of structural biology and made the biochemist's core skill set include cloning, protein expression, crystal growth, and synchrotron data collection.
Effect on the workStructural biology created a new subspecialty of biochemistry that commanded premium salaries in pharmaceutical drug discovery. Structure-based drug design (pioneered at companies like Agouron Pharmaceuticals, founded 1984) became a major driver of industrial biochemist hiring through the 1990s. The number of structural biology positions at pharmaceutical companies grew rapidly with each new synchrotron beamline and protein expressed.
Work toolChanging equipment NMR spectroscopy for protein structure in solution (Wuthrich method)
Kurt Wuthrich at ETH Zurich developed 2D NMR methods in the early 1980s that could determine protein structures in solution rather than requiring crystals. This was transformative: many biologically important proteins (membrane proteins, intrinsically disordered proteins, signaling proteins in their physiological states) do not crystallize well or at all. By the late 1980s, three-dimensional NMR with isotopic labeling could characterize proteins up to about 40 kDa in solution. Wuthrich received the 2002 Nobel Prize in Chemistry for this work. NMR created a second experimental pillar for structural biochemists and biophysicists and generated a new market for NMR spectroscopists at pharmaceutical companies and academic NMR centers.
Work toolChanging equipment Recombinant DNA expression, PCR, and genomics tools (molecular biology toolbox)
PCR (Mullis, 1983; Nobel 1993), recombinant expression systems (E. coli, baculovirus, mammalian cells), and the human genome project (completed 2003) gave biochemists an entirely new relationship with their starting materials. Before recombinant DNA, a biochemist studying an enzyme had to purify it from the native biological source, often kilograms of tissue. After, any gene could be expressed in bacteria and purified in milligram quantities in days. This compressed the bottleneck of biochemistry from protein isolation to experimental design. The genomics era then provided every protein sequence in every organism, creating the target landscape that structural and mechanistic biochemists would spend the next two decades characterizing.
Effect on the workThe molecular biology toolbox dramatically expanded what one PhD biochemist could accomplish in a year and increased industrial employment in pharmaceutical target identification and validation. Biotechnology industry employment of biochemists grew from near zero in 1980 to a substantial fraction of the profession by 2000, driven by Genentech, Amgen, Biogen, and a wave of startups using recombinant proteins as drugs.
Work toolChanging equipment Cryo-EM resolution revolution (direct electron detectors, 2013; Nobel Prize 2017)
In 2013, a new generation of direct electron detectors replaced the photographic film and CCD cameras used in cryo-electron microscopy, enabling the collection of movie-mode data that could correct for beam-induced sample motion in post-processing. The resolution of cryo-EM structures jumped from 5-8 Angstroms to 2-3 Angstroms within two years -- a leap that made cryo-EM competitive with and often superior to X-ray crystallography for large complexes. Jacques Dubochet, Joachim Frank, and Richard Henderson received the 2017 Nobel Prize in Chemistry for the foundational cryo-EM methods. By 2020, cryo-EM had become the primary method for determining structures of membrane proteins, ribosomes, viral capsids, and multi-megadalton assemblies that had resisted crystallization for decades. This opened structural biochemistry to a new class of biologically relevant targets and drove large investments in cryo-EM facilities at universities, national laboratories, and pharmaceutical companies.
Effect on the workThe cryo-EM revolution created strong demand for structural biologists with cryo-EM expertise and for the software engineers building the ML-accelerated data processing pipelines (CryoSPARC, RELION). Pharmaceutical companies built in-house cryo-EM facilities; academic cryo-EM centers expanded rapidly. A generation of biochemists trained in crystallography retrained or hired cryo-EM specialists.
Work toolChanging equipment AI structure prediction and generative protein design (AlphaFold 2 onward)
AlphaFold 2, released by DeepMind in July 2021, solved the 50-year-old protein folding problem for most single-chain proteins. AlphaFold 3 (May 2024) extended accurate structure prediction to protein-DNA, protein-RNA, and protein-small molecule co-complexes with approximately 80% success on novel drug-target interactions, making experimental structure determination optional for many hypothesis-testing applications. RFdiffusion (Baker Lab, July 2023) and ProteinMPNN (Baker Lab, 2022) together enabled generative protein design: a biochemist can now computationally design a novel enzyme or high-affinity molecular binder, generate hundreds of candidate structures in a day, and synthesize only the top-scoring ones for experimental validation. CryoSPARC v4 and ModelAngelo (DeepMind, 2024) automated the cryo-EM data processing and model-building steps that previously required weeks of manual work. ML interatomic potentials (MACE, NequIP) now run molecular dynamics at near-quantum accuracy 100-10,000x faster than traditional methods. These are not incremental improvements: they represent a categorical shift in what one biochemist can accomplish per unit of time, and they are the largest technology disruption in the profession's history since X-ray crystallography.
Effect on the workBLS projects 6% employment growth for 19-1021 through 2034 despite this disruption, reflecting that AI tools augment rather than replace the experimental judgment, mechanism elucidation, and instrument-level expertise at the core of the role. The pharmaceutical and AI drug discovery sector (EvolutionaryScale, Isomorphic Labs, Generate Biomedicines, Recursion) is creating new biochemist roles focused on designing and validating AI-generated molecular structures -- a job category that did not exist in 2020.
Work toolChanging equipment
What credible sources project
Scrub the slider past now to anchor each scenario on the scrubber. The spread is the range of futures credible sources project for this role.
What's shifting in the work right now
The historical view above shows how this role has moved. This is the present-day detail: which AI tools are picking up which tasks, where the edge still is, and the natural directions this work can grow.
What's changing in your day
Three parts of your work where AI is already doing real lifting, and what stays yours.
AI is sitting alongside you herePredict and interpret protein structures and biomolecular complex geometries using AlphaFold 3 and related structure prediction AI — submitting multi-chain jobs to the AlphaFold Server or local GPU deployment for protein-protein, protein-nucleic acid, and protein-small molecule complexes
Predict and interpret protein structures and biomolecular complex geometries using AlphaFold 3 and related structure prediction AI — submitting multi-chain jobs to the AlphaFold Server or local GPU deployment for protein-protein, protein-nucleic acid, and protein-small molecule complexes; interpreting pLDDT per-residue confidence scores and PAE (Predicted Aligned Error) inter-domain error matrices to assess which predicted regions are reliable enough for downstream structural analysis; validating high-confidence predictions against available PDB structures and experimental density; and using AI-predicted structures to guide mutational hypothesis generation, binding site identification, and comparative structural analysis.[3],[6],[1]
AlphaFold 3 now handles most single-protein and many complex structure prediction tasks that previously required months of experimental structural biology work. The residual human expertise that determines competitive advantage is interpretive: knowing how to read pLDDT and PAE matrices to identify which regions of a predicted structure are trustworthy versus speculative (intrinsically disordered regions, novel folds with few homologs, allosteric sites with high structural variability), and knowing which biologically relevant states AlphaFold is likely to miss (active vs. inactive conformations, ligand-induced conformational changes, co-crystal packing artifacts). Build deep fluency with structural interpretation tools (PyMOL, UCSF ChimeraX, Coot) so you can use AI-predicted structures as a starting point for mechanistic reasoning rather than treating them as ground truth.
AI is sitting alongside you hereDetermine three-dimensional structures of proteins and biomolecular complexes using AI-accelerated cryo-EM — preparing vitrified cryo-EM samples on carbon or graphene grids (including challenging membrane proteins in detergent or nanodisc reconstitution)
Determine three-dimensional structures of proteins and biomolecular complexes using AI-accelerated cryo-EM — preparing vitrified cryo-EM samples on carbon or graphene grids (including challenging membrane proteins in detergent or nanodisc reconstitution); collecting micrographs on a 200–300 kV transmission electron microscope; processing particle stacks in CryoSPARC v4 or RELION 5 using ML-guided 2D/3D classification to isolate conformational states; running ModelAngelo for automated atomic model building into density maps; and validating final models against MolProbity and PDB deposition standards.[9],[8],[1]
CryoSPARC v4's ML-accelerated heterogeneous reconstruction and ModelAngelo's automated model building have dramatically compressed the data processing timeline — a dataset that previously required weeks of manual classification and days of interactive model building now processes in days and hours. The persistent human bottleneck is sample preparation: achieving a vitrified cryo-EM grid that shows well-dispersed, correctly oriented particles with suitable ice thickness is a tacit skill that no software automates. Invest in mastering cryo-EM sample prep for your specific protein class (GPCRs, ribosomes, IDPs, membrane enzymes) — this physical technique skill determines whether AI-accelerated downstream processing can even run, and is far harder to replicate than the computational steps.
AI is sitting alongside you hereDesign novel proteins, enzymes, and molecular binders using generative protein design AI — specifying target function (enzymatic activity, binding affinity and selectivity, thermostability, expression yield) and using RFdiffusion to generate de novo protein backbones
Design novel proteins, enzymes, and molecular binders using generative protein design AI — specifying target function (enzymatic activity, binding affinity and selectivity, thermostability, expression yield) and using RFdiffusion to generate de novo protein backbones; applying ProteinMPNN to design optimal amino acid sequences for each generated backbone; filtering candidates with ESMFold or AlphaFold 2 structure prediction for sequence-backbone consistency; prioritizing candidates by predicted biophysical properties; and orchestrating the wet-lab synthesis, expression, and validation cycle to close the design-build-test loop.[4],[5],[7]
RFdiffusion + ProteinMPNN represent a paradigm shift in protein engineering: a biochemist can now generate hundreds of computationally validated novel protein designs in a day and then focus physical synthesis on the top-scoring handful. Your irreplaceable expertise is in specifying the fitness landscape — defining what "better" means in your biological context (which thermostability gain matters if it comes at the cost of reduced solubility in a cell; which binding affinity is sufficient given the assay format; which scaffold geometry is compatible with the downstream conjugation chemistry). The design cycle is also iterative: the biological context of each experimental round (why a design that looked good in silico expressed poorly, or bound but showed off-target reactivity) informs the next round in ways that require biochemical interpretation. Build a systematic design-test-learn protocol that captures this information explicitly across rounds.
Where this role is heading
Natural next steps for someone with your foundation: not exits, evolutions.
Natural Sciences Managers
Biochemists and biophysicists who develop research program management skills — directing multi-investigator research efforts, managing cross-functional teams of computational and wet-lab scientists, building CRO/industry partnerships, and leading AI tool strategy decisions for structural biology or protein design programs — transition naturally into Natural Sciences Manager roles at pharmaceutical companies, biotech organizations, or academic research centers. This path is especially timely in the current period: as AI tools (AlphaFold, RFdiffusion, CryoSPARC) rapidly transform structural biology workflows, organizations need research leaders who can evaluate which AI capabilities to adopt, set validation standards for AI-generated structural models and protein designs, and manage the human expertise transition. Natural Sciences Managers at pharma and biotech earn median $162,760 (BLS 2024), with BLS projecting +7% growth through 2034. The PI track record — grant management, publication leadership, trainee mentorship, vendor and CRO relationships — maps directly onto the management responsibilities required.
- · Research program management: portfolio prioritization across multiple structural and computational research threads, CRO and CXRO governance, milestone-based go/no-go frameworks for protein design and structure-function campaigns
- · AI tool strategy for structural biology: evaluating and deploying AlphaFold pipelines, cryo-EM automation systems, and protein design AI (RFdiffusion, ESM3) at organizational scale; setting validation standards for AI-generated structural claims used in regulatory submissions
- · People management in science: hiring computational biologists alongside wet-lab structural biologists, managing interdisciplinary team dynamics, conducting performance reviews for scientists with specialized technical skills
- · Business development skills: licensing deal structures for structural biology and protein design collaborations, SBIR/STTR grant management, sponsored research agreement (SRA) structures with pharma for structural characterization campaigns
- · Industry regulatory literacy: IND-enabling structural characterization data packages for biologics, FDA protein structure requirements for biosimilar submissions, ICH Q6B specifications for well-characterized biologics
See the same long-arc view for your own profession.
Browse the directory by industry, or search by title or SOC code. New roles ship every few weeks. Every profile cites every claim.
Browse all roles