The Orbion Blog
Protein Science, in Practice
Field-tested deep-dives on structural biology, protein engineering, expression, and the AI behind it all — written for the people at the bench
Protein Language Models vs MSA-Based Predictors: Which to Trust for Your Target
AlphaFold gives your orphan protein a soft, low-confidence fold after searching a shallow alignment. ESMFold returns a compact model in seconds with pLDDT…
Reconstituting a Membrane Protein into Nanodiscs: A Step-by-Step Troubleshooting Guide
You mix purified receptor, MSP1D1, and POPC at the ratio that worked in a paper. After overnight detergent removal, the tube is clear. The SEC trace is not:…
Solubility Tags for Difficult and Membrane Proteins: When MBP, SUMO, or Fh8 Actually Help
Your His-tagged construct is almost entirely in the pellet, so you replace His6 with MBP. The next expression test looks transformed: the fusion is abundant…
AlphaFold Says High Confidence, but Your Protein Misbehaves: Reconciling pLDDT with Bench Reality
Your AlphaFold model is blue from end to end. Median pLDDT is 92, the core looks compact, and the active-site residues sit exactly where the literature says…
Detergent Selection for Membrane Protein Extraction: A Decision Tree from DDM to LMNG
Your membrane prep contains the target. One percent DDM moves most of it into the supernatant. The yield looks excellent—until the sample reaches immobilized…
Rigid or Flexible? How to Choose the Linker Between Your Tag and Your Protein
Your fusion expresses beautifully. The gel shows a crisp band at the right size, the His-tag pulls it down cleanly, and yield is high. Then you run the…
How Many Homologs Does AlphaFold Need? MSA Depth, Orphan Proteins, and Single-Sequence Prediction
Your model came back at pLDDT 45 across almost the entire chain. No confident core, no floating disordered tail you can point to, just a uniform wash of…
GroEL Is in Your Prep: Why Chaperones Co-Purify With Your His-Tagged Protein
Your His-tagged protein came off the Ni-NTA looking clean, mostly. There is your band at the expected mass, and then there is that other one: a fat, stubborn…
N-Terminal or C-Terminal? Where to Put Your Tag (and Why It Changes Everything)
You designed two constructs of the same protein with the same His10 tag. The only difference: one carries the tag at the N-terminus, the other at the…
AlphaFold Gives You One Structure. Your Protein Has Several.
You typed in a sequence, waited a few minutes, and got back a single, beautiful model with a 90-plus pLDDT core. It looks definitive. It looks like the…
Why Won't My Protein Concentrate? Precipitation at High Concentration, Explained
Purification went flawlessly. The SEC trace is a single symmetric peak, the A280 says you have 8 mg spread across 12 mL, and all you need is 5 mg/mL for the…
Which PTMs Actually Survive in Your Expression Host? How to Read a PTM Prediction Before You Pick E. coli or HEK293
You ran your sequence through a PTM predictor and got back a clean list: two N-glycosylation sequons, a cluster of phosphosites in a cytoplasmic loop, three…
Your Disulfide-Rich Protein Won't Fold in E. coli: The Redox, Strain, and Periplasm Decision Guide
You subcloned your antibody fragment — or your growth factor, or your protease inhibitor — into a clean pET vector, transformed BL21(DE3), induced, and ran a…
ΔTm or ΔΔG? Which Stability Metric to Trust When You're Ranking Mutations
You ran a stability predictor on 40 candidate mutations, and now every row has two numbers. One mutation shows a predicted ΔTm of +6.2°C but a ΔΔG of only…
Your Protein Expressed as Inclusion Bodies — Refold, Redesign, or Re-Host? A Decision Guide
Your SDS-PAGE says the protein expressed beautifully — a fat band at exactly the right molecular weight. Your supernatant says otherwise: it's clear, and…
Live Now: Five Structure Models — and Which Mutations to Trust
Twice in the last month, a customer caught something wrong in our output before we did — a structure model quietly returning a bad fold, with everything…
AlphaFold Gave You a Structure, Not a Function: How to Predict What a Protein Actually Does From Sequence
Your AlphaFold model is beautiful — pLDDT 92 across the core, a clean two-domain architecture, a pocket you could practically dock a ligand into. And you…
Astra on Proteases: A 545-Protease Benchmark
Proteases are one of the most clinically validated enzyme classes in medicine — from the ACE and neprilysin inhibitors in cardiovascular disease to the HIV,…
Getting Your Protein Secreted: Signal Peptide Selection That Actually Works
Signal peptides are host-specific, cargo-dependent gatekeepers, not portable parts. How to match the leader to your expression host and secretion pathway…
Prediction-Guided FGF21 Construct Design: Our First Wet-Lab Study
Orbion's first public wet-lab study: a prospective, head-to-head test of prediction-guided FGF21 construct design. The AI kept the tail the textbook trims.
Calibration, Not Compounding: What Experimental Feedback Did for a Protein-Stability Model
Does feeding a protein-stability model your own lab results make it better on future proteins? Target-specific adaptation eventually helped the measured…