Hydrophobicity Plots Explained: The Kyte-Doolittle Scale and Finding Transmembrane Domains
6 min read · Updated July 10, 2026
A membrane protein's sequence alone can tell you a surprising amount about its structure, if you know how to read it. Hydrophobicity plots translate each residue's chemical character into a single score, then smooth those scores across a sliding window to reveal where a chain likely buries itself in a lipid bilayer versus where it sits exposed in solvent. This guide covers the Kyte-Doolittle scale, how window size shapes what you see, how to read a peak as a candidate transmembrane helix rather than a proven one, and where other scales like Hopp-Woods fit in.
What the Kyte-Doolittle scale measures
The Kyte-Doolittle scale, published in 1982, assigns each of the 20 amino acids a single hydrophobicity value built from water-vapor transfer free energies and the observed tendency of each residue to sit in a protein's interior versus its surface. Values run from about −4.5 for arginine, the most hydrophilic residue on the scale, up to +4.5 for isoleucine, the most hydrophobic.
Other strongly hydrophobic residues — leucine, phenylalanine, valine, methionine — cluster near the top of the scale; charged and polar residues — lysine, aspartate, glutamate, glutamine — sit near the bottom. A hydrophobicity plot works through a sequence one residue at a time, looking up each residue's value on this scale before any smoothing is applied.
Why the plot needs a sliding window
Plotted raw, per-residue values are too jagged to read: a single charged residue dropped into an otherwise hydrophobic stretch produces a sharp dip that has nothing to do with the protein's real topology. Hydropathy plots fix this by averaging the scale values over a sliding window of adjacent residues and plotting that average at the window's center position, which turns the noisy trace into readable peaks and troughs.
Window size is a genuine tradeoff, not just a display setting.
- A short window gives sharp, high-resolution output but stays noisy — a short run of hydrophobic residues can produce peaks that don't correspond to any real structural feature.
- A long window smooths the trace into a small number of dominant peaks but blurs exactly where a helix starts and ends, and can merge two adjacent short features into one broad hump.
- A window around 19-21 residues is the conventional default for whole-protein transmembrane screening, because it roughly matches the length of an alpha helix that spans a lipid bilayer.
Reading a peak as a transmembrane candidate
Kyte and Doolittle's original suggestion was that a sustained average score of about +1.6 to +2.0 over a roughly 19-residue window flags a likely transmembrane helix. The key word is sustained — you're looking for a stretch of the plot that stays above the threshold across roughly a helix's worth of residues, not a single-residue spike. A membrane protein typically shows a series of these peaks, one per predicted transmembrane segment, with the troughs between them corresponding to the connecting loops.
That said, a hydropathy plot predicts a physicochemical tendency, not a structural certainty. Treat any high-scoring region as a candidate transmembrane segment worth checking against known topology data or a dedicated transmembrane-prediction method, not as a confirmed structural assignment.
Kyte-Doolittle vs. Hopp-Woods: different scales for different questions
Kyte-Doolittle is built around one question: is this residue likely to be buried in a hydrophobic environment? Other scales are built around different questions. The Hopp-Woods scale, for example, is oriented toward predicting surface-exposed, antigenic regions — essentially the opposite emphasis, since a residue that's a good antibody epitope candidate is usually one that's accessible at the surface rather than buried in a core or a membrane.
Running both a Kyte-Doolittle-style hydropathy plot and thinking in terms of surface exposure gives you two complementary reads on the same sequence: where the hydrophobic core or membrane-spanning segments probably are, and where the solvent-exposed, more antigenic loops probably are.
Where this matters in the lab
In practice, a hydropathy plot earns its place early in a project, before you've committed to an expression or purification strategy.
- Membrane protein topology: a series of sustained peaks tells you roughly how many transmembrane helices to expect and where the connecting loops fall, which shapes decisions about tag placement, detergent choice, and construct boundaries.
- Surface loops versus buried core: troughs are better candidates for antibody epitopes or for inserting a tag without disrupting folding, while sustained peaks mark regions you'd generally avoid tagging or exposing.
- Solubility and aggregation risk: for a nominally soluble construct, unexpectedly long or frequent hydrophobic stretches are an early warning sign worth factoring into expression planning, even before you've expressed anything.
Plotting hydrophobicity on SeqBench
SeqBench's Hydrophobicity Plot tool runs this sliding-window calculation for you: paste a sequence and it scores every residue on the Kyte-Doolittle scale, then plots the windowed average so you can spot transmembrane and surface regions without building the calculation yourself.
Once you've flagged candidate transmembrane or surface regions, Protein Properties can compute the molecular weight, isoelectric point, extinction coefficient, and composition of the full sequence or a domain you've cut out, and Protease Digestion can show where trypsin, Lys-C, chymotrypsin, and other proteases would cut around those regions and what peptide masses to expect — useful when planning peptide mapping or MS work near a predicted membrane-spanning segment.
Frequently asked questions
What window size should I use for a hydropathy plot?
Around 19-21 residues is the standard default for scanning a whole protein for transmembrane helices, since it roughly matches the length of a membrane-spanning alpha helix. Use a shorter window if you need sharper resolution on a short motif, and expect a longer window to smooth out noise at the cost of blurring exactly where a helix starts and ends.
What Kyte-Doolittle score counts as a transmembrane helix?
Kyte and Doolittle's original guidance was a sustained average score of about +1.6 to +2.0 over a roughly 19-residue window. A single high-scoring residue isn't enough — the elevated score needs to hold across a stretch long enough to span a membrane.
Can a hydrophobicity plot alone predict transmembrane domains reliably?
Not on its own — a hydropathy plot predicts a physicochemical tendency, not a confirmed structure. Treat a high-scoring region as a candidate transmembrane segment and check it against known topology data or a dedicated transmembrane-prediction method before relying on it.
What's the difference between the Kyte-Doolittle and Hopp-Woods scales?
Kyte-Doolittle is built to find buried hydrophobic regions like transmembrane helices and core residues. Hopp-Woods is oriented the opposite way, toward predicting surface-exposed, antigenic regions, which makes it more useful for epitope prediction than for spotting membrane-spanning segments.
Related references
One-letter and three-letter amino acid codes with key properties.
Side-chain and terminal pKa values and how they set the isoelectric point.
Common affinity, epitope and solubility protein tags with amino acid sequences, mass, purification/detection and protease cleavage.