Files stay in your browser.
Windows .pdb debug-symbol files from Visual Studio are a different format and don't open here.
Type 1HSG into the search box above the canvas and press Enter. HIV-1 protease opens as a cartoon with its two chains in different colours, the inhibitor indinavir sits as sticks in the active site, and the crystal waters show as small dots. The panel beside the canvas fills with the entry's title, method, 2.00 Å resolution, both chains and the ligand, listed as MK1.
Click MK1 in that panel to select the inhibitor and frame it, then point at Representation, the top button in the toolbar on the left of the canvas, and pick Spheres. Only the inhibitor turns into spheres, because the style menus act on the selection while there is one. Click Clear above the selection and they act on everything again.
The PDB viewer is built on Mol*, the open-source engine the RCSB PDB and the AlphaFold Protein Structure Database use on their own sites, and it needs no account. Files you open from your computer are read in your browser and never uploaded. What you type into search goes to RCSB PDB, AlphaFold DB and PubChem, which send the structure straight to your browser.
The search box takes an ID or a compound name, and press / anywhere on the page to jump to it. Before you type, it offers a few examples under Try.
| You type | Example | What opens |
|---|---|---|
| PDB ID | 4HHB | The RCSB PDB entry, with its title, method and resolution shown before you open it |
| UniProt accession | P69905 | The AlphaFold DB prediction for that protein |
| AlphaFold DB model ID | AF-P69905-F1 | The same AlphaFold DB prediction |
| Compound name | caffeine | A list of PubChem compounds; the one you pick opens as its 3D conformer |
| PubChem CID | 60823 | That PubChem compound, atorvastatin here, in 3D |
Enter replaces whatever is open with the highlighted result. Shift+Enter adds it to the scene instead, which is how you put an AlphaFold model beside the crystal structure of the same protein. Escape closes the results, and pressing it again clears the box.
A PDB ID that RCSB does not hold is marked No RCSB PDB entry has this ID, and an accession without a model AlphaFold DB has no model for this accession, so neither loads by mistake. PubChem has no 3D conformer for some compounds, such as large or flexible ones; draw those in SketchIQ instead.
Drop files onto the canvas, or click the upload button at the right end of the search box. Select several files at once, such as a receptor and a docked ligand, and they open together in one scene, replacing what was there.
| Format | Extensions | Where these files usually come from |
|---|---|---|
| PDB | .pdb | The PDB archive, structure predictors, most modelling software |
| PDBx/mmCIF | .cif, .mmcif | The current PDB archive format, required for very large complexes |
| SDF and MOL | .sdf, .mol | Small molecules from PubChem, ChEMBL and docking tools |
| MOL2 | .mol2 | Ligands and receptors prepared for docking |
| PDBQT | .pdbqt | AutoDock Vina inputs and docked poses, read as PDB |
| XYZ | .xyz | Quantum chemistry output with plain Cartesian coordinates |
| GRO | .gro | GROMACS coordinate files from molecular dynamics |
| PyMOL session | .pse, .pze | Sessions saved in PyMOL, described under PyMOL sessions |
| Density map | .map, .mrc, .ccp4, also gzipped | Cryo-EM and crystallographic maps, described under Density maps |
| MD coordinates | .xtc, .dcd, .trr | GROMACS, NAMD and other simulations, opened over a topology as described under MD trajectories |
Predictions from AlphaFold 2, ESMFold, Boltz-2 and OpenFold3 open directly. Files must be plain text, so extract a .gz download first. Visual Studio also writes files ending in .pdb, but those are Windows debug-symbol files, not structures; the viewer recognizes them and says so instead of failing on them.
The panel beside the canvas describes each open structure. On narrower screens it sits below the canvas.
| Row | What it is |
|---|---|
| Title and source | The entry title, with a link to its RCSB PDB, AlphaFold DB or PubChem page. A file shows its name and Opened from your computer |
Method and Resolution | How the structure was solved, for PDB entries |
Mean pLDDT | The average confidence of an AlphaFold DB model, with its band, such as 98.1, very high |
Chains, Residues, Atoms, Waters | Counts for the whole structure |
Chains list | Each chain's ID, name and length. Click one to select and frame it |
Ligands list | Each ligand's three-letter code and name. Click one to select and frame it |
While something is selected, a Selection section at the top gives its residue and atom counts and its chains. Zoom to frames it again and Clear deselects it.
When you use one of the tools in the left toolbar, such as Ligand interactions or Mutate, its results take the top of the panel and the structure shrinks to a one-line header. Details opens the full description again and Less folds it. Tabs above the results start with Structure, followed by every tool you used in this scene that still has results, named as in the toolbar, so you can switch between, say, the contact table and the rotamer list.
The cross beside the tabs ends a tool: its previews and highlights go, while what you built, such as a mutant, measurements or a superposed copy, stays until you remove it from the tool's results. On a narrow screen, where the results sit below the canvas, a Results button on the canvas scrolls down to them.
| Action | Mouse or trackpad |
|---|---|
| Rotate | Drag |
| Move sideways | Right-drag, or Ctrl+drag |
| Zoom | Scroll, or pinch on a trackpad |
| Roll around the screen axis | Ctrl+Shift+drag |
| Clip the front of the scene | Shift+scroll |
| Focus on a residue or ligand | Click it |
| Fit everything back into view | Reset the view, in the toolbar at the bottom right of the canvas |
Clicking a residue or ligand zooms in on it and draws the residues within 5 Å as sticks. Hover over any atom and its chain, residue name and number appear in the corner of the canvas. To find a numbered residue, such as an active-site cysteine, the sequence view is quicker.
The toolbar on the left of the canvas has three groups. The first holds Representation, Color, Click to select, Lighting and Sequence; the tools below it are described in the following sections. Each opens a menu beside its button when you point at it, and a click pins it open until you close it or press Escape. A pressed button is the tool whose results the side panel shows, and a small dot means a tool has something drawn in 3D.
With nothing selected, Representation turns Cartoon, Surface, Sticks, Lines and Spheres on or off for everything, with a check beside each one shown. Structures open with cartoon and sticks on. Color recolours by Chain, Secondary structure, Element, B-factor, pLDDT or Hydrophobicity. pLDDT is meaningful for predicted models, which store confidence where crystal structures store B-factors.
The title of both menus says what they act on: All with nothing selected, or the selection, such as Selection, MK1 902 B. A representation is then added to the selection, a colour applies to it only, and Hide the selection appears in Representation to remove the selected atoms from view.
Selecting from the panel's chain and ligand lists is quickest. To pick by hand, open Click to select and choose Residues, Chains or Atoms; clicks in the 3D view then select instead of focusing, until you choose Off, a click focuses.
Lighting sets how the canvas is lit: Flat for standard lighting, Outline for flat colour with black outlines like an illustration, and Soft shadows for ambient occlusion in crevices. Images and movies are rendered in whichever lighting is on.
Sequence, in the first toolbar group, turns on a strip of one-letter residues along the top of the canvas, as in PyMOL and ChimeraX. On a narrow screen it sits above the command line instead. Every chain of every open structure is there in turn, with residue numbers every 10, and ligands and waters follow as one short group at the end. A dot between two residues marks a stretch the model leaves out, and hovering it says how many residues are missing. When more than one chain is open, the menu at the left of the strip shows a single chain instead of all of them.
Click a residue to select it, then shift-click another to add everything between them, or drag across a stretch to select it in one go. Ctrl+click, or Cmd+click on a Mac, adds a residue or takes it away, and clicking a chain's label selects the whole chain. The strip uses the same selection as the 3D view, so selected residues are green in both and the style, colour and analysis tools act on them. A selection made in 3D or with a command lights up in the strip and scrolls into view.
Pointing at a residue in the strip highlights it in 3D and names it in the corner of the canvas, and pointing at an atom in 3D outlines its residue in the strip. Scroll the strip sideways with the mouse wheel, or swipe it on a touch screen.
The strip stays on or off the next time you open the viewer in this browser. The PyMOL command set seq_view, 1 turns it on and set seq_view, 0 turns it off.
With 1HSG open, point at Ligand interactions, the first button of the second group in the left toolbar. It picks the first ligand, MK1 902 B, and with Show contacts in 3D on, dashed lines join indinavir to the residues it touches, which are labelled in the 3D view. The side panel lists the contacts under Ligand interactions: 9 hydrogen bonds, 1 salt bridge and 15 hydrophobic contacts, among them both catalytic aspartates, Asp25 of chain A and of chain B.
The table gives each contact's Residue, Type and distance in Å. Pointing at a row highlights it in 3D, and clicking it zooms there. The coloured kinds above the table, such as Hydrogen bond and Hydrophobic, switch each kind on or off in both the table and the 3D view. Download CSV saves the table.
2D diagram opens a flat drawing of the ligand with each contacting residue placed around it, its contacts drawn as lines in the same colours, with the distance printed on every line except the hydrophobic ones. Download SVG and Download PNG save it for a slide or a paper.
The viewer looks for hydrogen bonds, salt bridges, halogen bonds, metal coordination, π-stacking, cation-π and hydrophobic contacts with Mol*'s default geometry. A hydrogen bond needs a donor and acceptor within 3.5 Å (4.1 Å with sulfur) at a sensible angle; a salt bridge, charged groups within 5.0 Å; a hydrophobic contact, two carbons within 4.0 Å. Water-mediated contacts and weak C-H···O bonds are not listed.
For a docked pose, open the receptor and the pose file together. A pose opened alone has no protein to contact, and the panel says so. The button is greyed out when no open structure has a small-molecule ligand. From the search box or the command line, interactions MK1 902 B picks a ligand and interactions off hides the lines; closing the results hides them too.
Binding pocket, the target button in the left toolbar, shows which residues line a ligand's site. Click it, pick the ligand under Ligand and set Residues within, from 3 to 12 Å and 5 Å by default; the pocket residues turn to sticks with amber carbons. Pointing at the button only previews the menu, and the pocket is drawn once you click it. With indinavir in 1HSG, 33 residues lie within 5 Å, from both flaps and both catalytic aspartates. Focus on ligand frames it.
The side panel lists the pocket under Binding pocket, each residue with its Closest, Å distance to the ligand. Pointing at a row highlights the residue and clicking zooms to it. Copy residue IDs copies the list in the A:25,B:50 form that AutoDock Vina takes for flexible residues; some browsers block the clipboard, and the viewer then says so. Closing the results removes the highlight.
Docking results usually come as one file with many poses. Open the receptor and a multi-pose SDF or PDBQT file together, and the panel lists every pose with its name and score, read from SDF data items such as minimizedAffinity and CNNscore or from PDBQT REMARK VINA RESULT lines. A header such as Pose 3 of 3 · Affinity (kcal/mol) says which one is shown.
The arrows beside it, the arrow keys or a click on a row change the pose, and the pocket follows: in a test with three indinavir poses shifted against each other, it went from 33 residues to 31. Overlay all poses draws the other poses as faded grey sticks. A multi-model file without scores, such as an NMR ensemble, lists its models the same way.
Measure, the ruler button in the left toolbar, measures between atoms you click. Choose Distance, Angle or Dihedral, and a prompt above the structure asks for each atom in turn, such as Pick atom 1 of 2, while the cursor turns to a crosshair. Numbered labels mark the picks in order, and the menu counts them, as in 1 of 2 atoms picked. Clicking a bond takes its nearer atom and clicking a cartoon takes the residue's Cα.
Clicking the last atom again takes it back, Cancel drops the picks, and Escape closes the menu. While you pick, a click no longer zooms the camera.
Polar finds nitrogen and oxygen pairs 2.5 to 3.6 Å apart that can hydrogen-bond, including waters and ligands, and draws them as yellow dashes. With Everything it searches the whole scene: haemoglobin, 4HHB, has 1,946 such contacts. With Selection it finds the contacts that the selected atoms make with any atom, such as the residues and waters holding a haem. Each structure keeps at most 2,000 contacts, the closest ones.
Measurements are listed under Measure in the side panel, together with any made by the distance command. Pointing at one highlights its atoms and clicking zooms to it; the cross deletes it. Copy as CSV copies the list and Remove from 3D removes every measurement.
Superpose fits one structure, chain or ligand onto another. Choose a method, pick what to move under Move and what to fit it to under Onto, and click Superpose. The side panel then shows the RMSD, the TM-score, how many residues or atoms were Aligned and the Method.
| Method | Use it for |
|---|---|
TM-align | Structures with little sequence identity; it matches by shape |
Sequence | Close homologs and models of the same protein; it aligns the sequences, then fits Cα atoms and rejects outliers, as PyMOL's align does |
Ligand | Two copies of a ligand in different structures; it fits matching atoms by name, or by their largest common substructure |
With one structure open, the pickers offer its chains. A chain moved onto another chain of its own structure is copied into a new structure first, so the original stays where it was, and the button then reads Superpose a copy. Haemoglobin's chain B onto chain A with TM-align gives an RMSD of 1.51 Å over 139 residues and a TM-score of 0.897.
Remove copy deletes the copy, and for a whole structure Move back returns it to where it was. The PyMOL commands align, super and cealign do the same from the command line.
Type mutate A/82 PHE into the search box or the command line and press Enter. The viewer zooms to valine 82 of HIV-1 protease, draws the phenylalanine side chain that clashes least, in pink, and opens Mutate in the side panel. It lists all four phenylalanine rotamers under Rotamers, fewest clashes first: m-80 places with 0 clashes, while t80, despite being the second most common rotamer in the library at 34.3%, runs into 8 atoms.
The residue can be written as A/82, A:82 or just 82, and the amino acid as a one-letter or three-letter code.
You can also start from the toolbar. Click Mutate, the flask button, and a prompt above the structure says Pick a residue; Escape cancels. Click one residue in the 3D view and the menu shows a grid of three-letter codes in three groups, from hydrophobic to charged, with the current one outlined. Pick the new one.
The rotamers appear in the side panel, each row with its Frequency in the library and its Clashes. Previous, Next or the arrow keys step through them, and red discs in the canvas mark each clash of the one shown, with the worst overlap named under the list. The menu shows the rotamer on screen and its clash count beside Apply V82F, labelled with the mutation, which builds the mutant.
The mutant opens as a new structure, such as 1HSG mutant, with the mutated residues drawn as sticks, the original's colours and the view unchanged, and the original is hidden.
Further mutations add to the same mutant and are listed under Applied to the mutant, each with its rotamer and clash count. The cross beside one removes it, and Undo takes back the last; undoing every mutation shows the original again. Only one structure is mutated at a time, so export the mutant if you want to keep it and undo its mutations before mutating another structure.
Export PDB downloads the mutant, with header remarks naming each mutation, its rotamer and its clash count. Relax or refold opens Send to tool with the same file, for example to relax it by molecular dynamics in GROMACS.
Treat the result as a starting model. Side chains are placed from the Top8000 rotamer library on the fixed backbone, with no energy minimisation, so neither the backbone nor the neighbouring side chains move to make room. A clash is a heavy-atom overlap of 0.4 Å or more; hydrogens are not modelled.
Placed atoms carry a B-factor of 99, so they stand out when you colour by B-factor. Glycine and alanine have no rotamers to choose from, and a residue that is not a standard amino acid or lacks N, CA or C cannot be mutated.
With 1HSG open, point at Electrostatics, the lightning button in the left toolbar. The menu offers three sources of potential, Estimate, APBS and .dx file, each with one button. On Estimate, click Colour surface.
In a couple of seconds the protease gets a molecular surface coloured from red through white to blue, and the side panel reports a Net charge of +6 e from 70 charged atoms. Red is negative potential, blue positive, and the legend under Colour range gives the scale in kT/e.
This estimate is free and runs in your browser, but it is an approximation, as the information icon beside Coulombic estimate in the results says. It works like ChimeraX's coulombic: point charges in a distance-dependent dielectric of 4r, read 1.4 Å out from the surface. Unlike ChimeraX, it places formal charges on the ionizable groups at pH 7 instead of partial charges on every atom, so it needs no hydrogens, and it has no solvent or ion screening.
Use it to see where the charged patches are, not for figures you publish. Download .dx, in the side panel, saves its grid as an OpenDX file.
For a Poisson-Boltzmann result, choose APBS and click Open in PDB2PQR · 10 credits. It opens PDB2PQR with the structure as shown, already loaded as a PDB file; turn on Run APBS electrostatics there. Nothing runs or is charged until you submit the job yourself. A structure too large for PDB format cannot be handed over this way; download it and upload it to PDB2PQR instead.
When you come back to the viewer in the same tab, the APBS choice says Sent 1HSG to PDB2PQR and offers Open .dx result for the potential the job returned. .dx file opens an OpenDX potential from APBS, PDB2PQR or other software with Open .dx file; Colour surface then colours the surface with it, adding a molecular surface if the structure has none.
Maps open from this menu only; dropping a .dx file on the canvas is refused. A map that does not overlap the structure gets a warning, since it was probably computed for something else, and Open another adds a second map to choose between.
Colour range sets the potential at which the colours saturate, from ±0.5 to ±30 kT/e. Maps start at ±5 kT/e, the usual range for APBS figures. The Coulombic estimate starts at ±10 kT/e, because unscreened formal charges give larger values.
While a surface is coloured, the menu names the source on a Coloured by line, such as Coloured by Estimate. Remove from 3D, there or under the results, puts back the surface's own colours and removes the surfaces the tool added, and picking a scheme from Color ends the electrostatic colouring too.
From the search box or the command line, electrostatics coulombic, electrostatics map and electrostatics off do the same. The Coulombic colouring is kept when you reload the page; a .dx map, like a file from your computer, is left out of share links.
With 4HHB open, click Surface area, the droplet button in the left toolbar. On Surface, pick the Whole structure, one chain or the current Selection under Scope, and click Calculate. The side panel gives the solvent accessible surface area (SASA) in Ų, in total and for each chain, then a table of residues with their area, their relative SASA in percent and whether they are Buried, Intermediate or Exposed. Click a column heading to sort, click a residue to zoom to it, and Download CSV saves the table.
Relative SASA divides a residue's area by the largest area that amino acid reaches in a Gly-X-Gly peptide, from Tien et al. (2013). Below 20% counts as buried and above 50% as exposed. Ligands and the first and last residue of each chain have no relative value. Once a result is in, Colour by relative SASA colours the structure from blue for buried to orange for exposed, with everything that has no value in grey, and Remove from 3D puts the earlier colours back.
Each chain is measured with the rest of the structure in place, ligands and ions included, so chain A of 4HHB has 5,666.4 Ų in the tetramer and 7,904.2 Ų on its own. Water and hydrogens are left out.
Interface tells you how much surface two partners bury when they bind. Pick two chains, or a chain and a ligand, under Partner A and Partner B and click Calculate. The viewer measures each partner alone and the two together, and reports SASA(A) + SASA(B) − SASA(AB) as the buried surface area. For chains A and B of 4HHB that is 1,616.9 Ų, 804.1 Ų from chain A and 812.8 Ų from chain B, so halve the total for the area one side loses. The table lists every residue that loses area when the partner is added, with its area alone, in the complex and buried, and Colour interface residues paints them in one colour per partner.
The numbers come from the Shrake-Rupley method with a 1.4 Šprobe, 92 points per atom and Bondi radii, the same as the SASA calculator, so 1CRN gives 2,983.99 Ų in both. FreeSASA's default Lee-Richards method with ProtOr radii gives 2,999.32 Ų for 1CRN, 0.5% more, and 1,651.8 Ų for the 4HHB interface, 2% more. Compare structures measured with one method.
In the command line, get_area prints the total for a selection, as PyMOL does with dot_solvent on, for example get_area chain A and polymer.
Density map shows a cryo-EM or crystallographic map over the model. On the EMDB tab, type an ID such as EMD-33198 and click Load. On the File tab, Open map file takes a CCP4, MRC or MAP file of up to 256 MB, gzipped or not; map files can also be dropped on the canvas. With 7XHT open, emdb EMD-33198 in the command line does the same as the ID box.
EMDB maps come from the PDBe density server, downsampled, and open at EMDB's recommended contour, shown in σ: for EMD-33198 that is 8.46 σ, or 0.527 in the map's own units. A map file opens at the level that the top 1% of its values exceed, which hides most solvent noise in cryo-EM maps.
The menu sets the Contour level in σ or Absolute units, draws the map as a Surface or a Mesh, and sets its Colour and Opacity. Around selection, with a Within distance, shows the map only near the selected atoms, like PyMOL's carve, while σ still describes the whole map. Close map removes it.
The side panel lists each map's Source, Grid, Voxel size, Mean, σ, Range, Contour and what is Shown. EMDB maps come back when you reload the page and travel in share links. A map file does not: after a reload the viewer asks you to open it again.
MD trajectory plays a simulation over its topology, in two steps. First open the topology as a PDB or GRO file, the usual way; the menu then names it on a Topology line. Then click Open coordinates for the XTC, DCD or TRR file. You can also drop both files on the canvas together, and coordinates opened first wait for a topology with the same number of atoms. The limits are 150 MB and 25 million atom-frames, the number of atoms times the number of frames.
A player appears along the bottom of the canvas to play and pause, step a frame at a time, drag to a frame, change speed and loop.
The side panel shows the file, the number of frames and atoms and the time span, and plots the RMSD of every frame against the RMSD reference frame set in the menu, the first by default; Use frame takes the one on screen. RMSD is measured over the selected atoms, or every Cα when nothing is selected. Clicking the plot jumps to that frame, and the values can be saved as CSV.
The coordinates stay in your browser's memory, not in the tab's autosave. After a reload the topology comes back as a single frame and the panel says Not in this tab; opening the same coordinate file again with Open other coordinates brings the frames back with your styles.
Export, at the top right or as the download button in the left toolbar, has a tab for each kind of file. Image, Movie and 3D model export the view as it is, so set the camera and styles first; Structure saves the atoms themselves.
Image saves a PNG in the lighting set under Lighting. Size is 1×, 2× or 4× the canvas, with the pixel size shown underneath, or Custom in pixels; a size larger than your graphics card can render is reduced and marked. Background is Transparent, White or Black.
Download PNG saves the image and Copy puts it on the clipboard. Some browsers refuse a page's request to copy an image; the viewer then says The browser blocked copying the image. Download the PNG instead.
Movie renders an MP4 at 30 frames per second in the current lighting. Motion is Rock, which sways 20° each way every 4 seconds, Spin around the vertical axis or Rotate top over bottom. Duration runs from 2 to 20 seconds, Resolution is 1280 × 720, 1920 × 1080 or 1080 × 1080, and the background is white or black. Render MP4 takes about as long as the movie or longer and can be cancelled.
3D model downloads what is shown as a mesh: STL for slicers and 3D printing, OBJ with its colours in a zip with its material file, or glTF as one .glb file for Blender, the web and AR. Surfaces and spheres print as solid pieces, while cartoons and thin sticks snap without supports.
Structure writes PDB, mmCIF or, for the ligands and ions alone, SDF. All visible takes what is drawn and leaves out hidden chains and hidden structures, One structure takes one from the list, and Selection takes the selected atoms. A superposed structure is saved where it was moved to, and a mutant keeps the header remarks naming its mutations. The file is named after the entry, such as 4hhb.cif or 4hhb_selection.pdb.
Several structures go into one file as chains of a single model rather than as separate models, since most programs read only the first model of a PDB file. When two structures share a chain letter, the later one moves to the next free letter, and both the file and the message after saving say which. PDB holds at most 99,999 atoms and one-character chain IDs, so for a ribosome such as 4V6X the viewer explains why and asks for mmCIF instead.
From the command line, PyMOL's png saves the view too. png figure.png, 10cm, dpi=300 gives a 10 cm wide image at 300 dpi. Sizes are in pixels unless they carry cm, mm or in, which convert at 300 dpi when no dpi is given, and a missing side keeps the canvas's proportions. ray and quiet are accepted and change nothing, since every image is rendered off screen anyway.
Send to, the paper-plane button near the bottom of the left toolbar, hands the open structure to the tool that usually comes next. The tool opens with the files filled in and its price shown, and nothing runs or is charged until you submit it there.
| Row | Opens | What it receives |
|---|---|---|
Fix structure | PDBFixer, 3 credits | The whole structure as PDB |
Dock into this pocket | AutoDock Vina, 15 credits | The protein chains with only the picked ligand, so Vina's automatic box centres on that pocket |
Predict complex | Boltz-2, 50 credits | One FASTA file per protein chain and the ligand as SDF |
Open ligand in SketchIQ | SketchIQ, free | The ligand as a 2D drawing, ready to edit |
The ligand is the one named at the top of the menu, the same one Binding pocket and Ligand interactions use. With no ligand open, the Vina row reads Dock a ligand and sends the receptor alone, so you add the ligand and set the box in Vina. Sending MK1 from 1HSG to SketchIQ opens indinavir, C36H47N5O4, with its name and properties. Other tools… opens the full list of ProteinIQ tools to send the structure to.
The two arrows at the top right, beside Reset, undo and redo changes to the scene: styles and colours, hiding and showing, structures, maps and trajectories you open or remove, superpositions, measurements, mutations, electrostatic colouring and commands typed in the command line. Ctrl+Z undoes, Ctrl+Shift+Z or Ctrl+Y redoes (⌘Z and ⇧⌘Z on a Mac), and undo and redo work as commands too. Hovering over an arrow shows what it will take back, and a short message names each change as it goes, such as Undid: color by chain.
Undo leaves the camera, the selection, the lighting, the trajectory frame or docked pose on screen, and the binding pocket and interaction overlays as they are. History lasts while the tab is open and starts again after Reset, Start fresh, a reload, or opening a structure in place of the scene, with Enter in the search box or by dropping a file.
Reset, at the top right, takes the scene back to the structures as they were first opened, with their original styles and view and without measurements, mutants or overlays. Start fresh clears the scene, every tool and the tab's saved state, and offers to undo it for a few seconds. In the command line, reset scene and reinitialize do the same, while reset on its own only resets the view, as in PyMOL.
The viewer saves the scene in each browser tab as you work, with its structures, styles and camera, so reloading the page brings it back. Two tabs keep separate scenes, and closing a tab discards its scene.
A very large scene can exceed what the browser keeps for a tab; the viewer then says This scene is too large to keep across reloads in this tab. If a saved scene cannot be restored, or takes longer than half a minute, the viewer gives you the canvas back with a message offering Start fresh.
Export > Link copies a link that opens the same scene, styles and camera for anyone. Structures in it are fetched again from RCSB PDB, AlphaFold DB or PubChem when the link opens.
Files opened from your computer stay in your browser, so share links can't include them, and the same goes for mutants and PyMOL sessions, which are built in your browser: the link leaves them out and says so, and a scene made only of your own files cannot be shared. The scene is stored in the part of the link after #, which browsers do not send to ProteinIQ's server.
PyMOL users can type the commands they already know. In the search box, a line whose first word is a command and that has arguments, such as fetch 1hsg or show sticks, organic, runs as a command; starting a line with > forces it.
Command line, at the bottom of the canvas, opens a full command line with a log of recent commands and their results; commands run from the search box are logged there too. The up and down arrows recall earlier commands, which are remembered in this browser, and Tab completes command names, object and selection names, representations, colours and settings. Escape folds it back into its button.
| Command | What it does |
|---|---|
select name, sel, deselect | Stores a named selection and highlights it; one argument names it sele |
show, hide, as | Representations: cartoon, ribbon, sticks, lines, spheres, nb_spheres, nonbonded, surface, mesh, dots, putty, ellipsoids, labels, everything |
color, set_color, bg_color | PyMOL colour names, 0xRRGGBB, #RRGGBB, [r, g, b] or atomic |
util.cbc, util.cbag and the other util.cba* | Colour by chain, or carbons in one colour with other elements by type |
spectrum b, spectrum count | B-factor or residue-number gradients, for whole objects only |
set transparency, cartoon_transparency, stick_transparency, sphere_transparency | Transparency, optionally for a selection |
set seq_view, 1 or 0 | Shows or hides the sequence strip |
zoom, orient, center, reset | Camera |
align mobile, target | Sequence alignment, then a Cα fit with up to 5 outlier-rejection cycles, as in PyMOL; prints RMSD before and after |
super, cealign | Structural superposition with TM-align; cealign keeps PyMOL's argument order but does not run CE |
distance | One distance between two atoms, or every pair within a cutoff (4 Å by default), up to 100 drawn |
label sel, expr | Labels such as resn + resi, on up to 300 atoms; expressions are read, never executed |
remove, delete, enable, disable | Hide atoms and leave them out of later selections, delete objects and selections, show or hide objects |
fetch, load | PDB IDs, UniProt accessions and AlphaFold DB IDs, or an https link to a .cif, .bcif or .pdb file |
png | Saves the view as a PNG, at a size in pixels or in cm, mm or in at a given dpi |
save file, sel | Saves a .pdb, .cif or .sdf file of the selection, or of everything shown, as save 4hhb_a.pdb, chain A |
get_area sel | Solvent accessible surface area of the selection in Ų, with the rest of its structure in place |
count_atoms, get_names, help | Information |
Separate several commands on one line with ;. Lines copied from a Python script, such as cmd.color("red", "chain A"), run too. Selections follow PyMOL's syntax, including object names (1hsg and chain A), named selections and slash notation (/1hsg//A/25/CA).
Commands with no equivalent here yet, such as turn, alter and h_add, answer not supported yet: turn, and so does a first word the viewer does not know.
set works only for the settings above, and distance does not find polar contacts (mode=2). The viewer's own commands, mutate, interactions and electrostatics, run alongside the PyMOL ones. Named selections last only until you reload the page, and share links do not include them.
Drop a .pse or .pze file onto the canvas, or open it with the upload button, and its molecules load beside whatever is already open. The file is read in your browser, and it is only read as data, so a session file cannot run code. The viewer brings over each molecular object with its enabled state, representations, per-atom colours and labels, and says what it left out, for example Not opened: density_map (map).
Maps, meshes, CGO objects, measurements, selections, the saved camera view and object matrices are not carried over, and multi-state objects open in their first state only. A molecule with more than 64 colours keeps the 64 most common, and only the first 300 labels are shown. A session saved with pse_binary_dump on cannot be read; in PyMOL, run set pse_binary_dump, 0 and save it again.
A viewer shows a model as the file describes it, mistakes included. Before using a structure for docking or simulation, run it through the tools that measure quality:
Rendering runs on your device's graphics card through WebGL. Structures with a few hundred thousand atoms, such as ribosomes or viral capsids, load slowly on laptops and phones. The largest PDB entries come only as PDBx/mmCIF, which the viewer reads directly.
It draws structures with Mol*, the open-source engine behind the RCSB PDB and AlphaFold DB websites, on your device's graphics card through WebGL. Files you open are read in your browser and never uploaded; only what you type into search goes to RCSB PDB, AlphaFold DB or PubChem to fetch a structure.
Drop the .pdb file onto the canvas, or click the upload button at the right end of the search box. To open an entry from the archive instead, type its PDB ID, such as 4HHB, into the search box and press Enter.
PDB, PDBx/mmCIF, SDF, MOL, MOL2, PDBQT, XYZ and GRO structure files, PyMOL sessions saved as .pse or .pze, CCP4 and MRC density maps, and XTC, DCD or TRR trajectories over their topology. The search box also opens AlphaFold DB models by UniProt accession and PubChem compounds in 3D by name or CID.
Yes. Export > Link copies a link that opens the same structures, styles and camera view. Structures opened by PDB ID, UniProt accession or compound name are fetched again from their database. Files from your computer, PyMOL sessions and mutants stay in your browser, so links leave them out.
Yes. Drop or select several files at once to open them together, or press Shift+Enter in the search box to add an entry to the structures already open, for example an AlphaFold model beside the crystal structure of the same protein.
For looking at structures, styling them, inspecting a binding site and making figures, yes, and it needs no licence or account. It runs common PyMOL commands such as select, show, color, align, distance and png, opens .pse sessions, mutates residues with a rotamer library, saves PDB, mmCIF and SDF files, and exports images, movies and 3D models. It does not run Python scripts or PyMOL's ray tracer, and it does not carry over maps, multi-state objects or the saved view from PyMOL sessions, so keep PyMOL for those.
No. The viewer runs in a current version of Chrome, Edge, Firefox or Safari with WebGL turned on, as it is by default. Very large structures, such as ribosomes, load slowly on laptops and phones.
Yes. Drop it onto the canvas and its molecules open with their representations, colours and labels. Maps, measurements, selections and the camera view are not carried over. A session saved with pse_binary_dump on has to be saved again in PyMOL after set pse_binary_dump, 0.