Inside the automated lab.

It means 200,000 spots, one sequence each. A peptide shown in duplicate takes two spots, so replicates share the capacity: 100,000 different sequences in duplicate fill the same microarray.

Synthesized on it. Each peptide is built residue by residue on the coating of the slide, in situ, instead of being made separately and spotted. All peptides of the library are built in the same run.

From the C-terminus to the N-terminus, by conventional Fmoc chemistry. The C-terminus is bound to the surface. The N-terminus is the free end, where a label or reporter goes when the design has one.

A three-residue spacer, β-alanine–aspartate–β-alanine, at the C-terminus of every peptide. On the 3D coating it sits on polymer chains that already hold the peptide away from the glass. The spacer offsets the charge of the amine-bearing surface and reduces non-specific binding of the protein to it. You can extend it or replace it with a spacer of your own.

The 3D coating carries about 2.5 nmol of anchor groups per cm². Not every anchor yields a peptide the target protein can reach. Depending on the size of the protein, the accessible density can be below 0.25 nmol per cm².

Yes, per sequence. The share of anchor groups that carries the peptide is set to 100, 50, 25, 10, 5 or 2 %. A dilution series of one sequence shows whether its signal depends on density, which points to avidity rather than 1:1 binding.

Yes. Any commercially available Fmoc-protected amino acid with a free carboxyl group, sourced per project, including D-amino acids. Cyclization by thioether, disulfide or amide. Up to 25 residues. Stabilizing modifications are tested in the same run as the unmodified sequence.

26 in total, split freely between standard and custom building blocks, including any spacer or reporter the project needs. Most projects need no reporter, and the spacer can be adjusted. A library that uses all 20 standard amino acids therefore has room for up to six custom ones.

An Fmoc-protected amino group, a free carboxyl group, and an acid-labile protecting group, such as tBu, Trt or Boc, on any side chain that could react during synthesis.

Up to 25 residues, plus the spacer. Synthesis yield depends on the sequence more than on the length alone. For long or difficult sequences, a reporter at the N-terminus shows the relative yield of each peptide (step 1.5).

Rarely. Glycosylated building blocks are bulky and couple with low yield. Where one is available with Fmoc and side-chain protection, it can be tried as a custom building block.

By default, yes. The N-terminus of every peptide is acetylated, and the C-terminus is bound to the surface through the spacer. Peptides are linear unless the design asks for cyclization.

Yes. Peptides to be acetylated are marked in the sequence list by an N-terminal [Ac] residue, and the acetyl group then counts as one residue toward the peptide length. Acetylating the whole library is a single capping step instead and does not count toward the peptide length.

Yes, at the free N-terminus: biotin, desthiobiotin or a fluorophore, coupled as a custom building block. A reporter at the end of the chain makes the relative synthesis yield of each peptide visible. In Protease Profiling, its loss marks the cut (step 3.4).

No, it tends to reveal more. A ring loses less conformational entropy on binding than a flexible linear peptide, so it binds more strongly, and weaker binders become visible. In one K_D Ranking project, 78 of 99 cores that bound did so only as the cyclic peptide.

Not at present. Ring closures that need an external reagent, such as ring-closing metathesis or a chemical clip, are not offered. Thioether, disulfide and amide rings are.

Yes. Partial cyclization leaves each spot with a mix of linear and cyclized copies of the same sequence.

Yes. Head-to-tail amide cyclization needs one extra building block at the C-terminus. It takes one of the 26 building-block slots and adds one residue to the ring.

Synthesis controls, read with an antibody stain, in a fixed frame and grid in every block, and blank spots without peptide. Controls for your assay, positive or negative, are part of your library (step 4.3).

Yes. The project report includes the quality control of every microarray: the synthesis controls and the uniformity of the signal across the slide.

For binding assays, yes: a blocking buffer covers the surface before the protein is added. For Protease Profiling, no: the protease meets the peptides in its own buffer.

In binding assays, the reference scan shows what binds without the target protein: the detection reagent alone, on the peptides and on the coating. Each peptide's signal after incubation is read against it, so a peptide that binds the detection reagent is not mistaken for a binder of the protein. Protease Profiling uses a buffer-only well as its reference instead.

Red, around 635 nm, whenever the design allows, because peptides and coating fluoresce least there. Green, at 532 nm, serves a second readout in the same incubation, for example IgM in green next to IgG in red. Blue, at 488 nm, is a third, with the highest background.

No. The scanner reads fluorescence, not chemiluminescence. Use a secondary antibody labeled with a fluorophore that is read in the red channel.

Barely in the red channel. In the green and blue channels, aromatic residues add a noisy background: tryptophan most, then tyrosine and phenylalanine.

No. Send it labeled with a fluorophore or biotin, tagged for detection with a secondary antibody, or let us label it. Amount and buffer are confirmed in the quote.

Biotin, including a biotinylated Avi-tag, read with labeled streptavidin. HA, FLAG and Myc tags, read with labeled anti-tag antibodies. A mono-biotinylated protein is the most common format. A truncated construct needs the same tag as the full-length protein.

No, but it is recommended. K_D Ranking also works with a tagged protein and a labeled secondary reagent after the incubation. During that staining step, weakly bound protein can dissociate from the surface. A direct fluorophore label skips the step, and more of the weak binders stay visible.

Its molecular weight and extinction coefficient. We can usually derive the coefficient from the sequence. The formulation should contain no other protein, such as BSA; we confirm the buffer with you. A certificate of analysis helps but is not required.

A dye read in the red channel, around 635 nm. Charged dyes and excess dye cause non-specific binding to the peptides, so we agree the dye and the degree of labeling with you before the protein is labeled.

Not reliably. On a peptide array, anti-His detection gives a background that masks the protein's binding. Use biotin, an HA, FLAG or Myc tag, or a direct label instead.

PBS with Tween 20 (PBST), incubated overnight at 4 °C. If you have conditions of your own, we use them instead, which also keeps the results comparable with earlier assays.

Yes. Conditions from SPR, TR-FRET or similar methods can be reproduced on the microarray. HBS also works as an incubation buffer.

The one it is most stable in, where it is least prone to aggregation. Glycerol, sugars and surfactants do not affect the assay, and the formulation does not affect the microarray. If we label the protein, see step 2.4 for what to leave out.

All three. A purified protease in its buffer gives its specificity motif. Plasma or serum gives the stability that matters for a therapeutic candidate. Buffer series across pH show where an enzyme is active, and let you emulate the compartment your peptide must survive. One microarray holds up to 18 conditions.

A single condition over the whole slide takes 2 mL. Two wells take 200 µL each, six wells 150 µL, 12 wells 100 µL and 18 wells 80 µL. Each well is prepared with 10 µL extra for pipetting.

It depends on the weakest affinity you want to resolve. To saturate the surface-bound peptides, the highest concentration of the series must lie well above that K_D. Each well needs its concentration times its volume, summed over the series. The sample calculator in Learn (step 2.6) does the arithmetic. Your stock must be at least as concentrated as the highest well. If sample is limited, the series can be scaled down, at the cost of resolving only stronger binders.

Send at least 20 µg of each antibody, or 50 µg if sample is not limited. The sample calculator in Learn (step 2.6) shows the amount per well for any concentration, and the quote confirms the one for your project.

Frozen on dry ice in most cases. IgGs and nanobodies can also travel at 4 °C. Follow the supplier's recommendation for your protein, and use an express courier.

Yes. The analysis is a fixed, deterministic program: the same scans and library give the same numbers. Each delivered run is archived with its program version and settings, so it can be rerun later with identical results.

Through the grid. Control spots in a fixed frame and grid mark every block, and the image is aligned on them. Each spot's position then names its peptide. Usually the protein's own signal on the controls is enough for alignment. Otherwise the controls are stained separately.

They are flagged and left out of the peptide's value. They stay in the spot data you receive, so every exclusion can be checked.

Each peptide is compared with blank spots and with a scan taken before the target protein is added. It is called bound when its replicates lie significantly above that background, reported as a z-score and a false-discovery rate. The fold over background is reported with it.

A peptide whose signal is above background, but whose call is less certain than a binder's by the false-discovery rate. Candidates are listed apart from the binders, not mixed in with them.

Then the peptide is a secondary binder. It binds the staining reagent, not the target protein, and is kept out of the calls for the protein. The reference scan identifies it. In a titration, its signal does not rise with the protein concentration.

The analysis tests that. Binding is correlated with peptide properties across the library: charge, hydrophobicity and aromatic content. A strong correlation points to binding driven by those properties, a weak one to sequence-specific binding.

Not reliably. An end-point signal depends on synthesis yield and off-rate as much as on affinity. To rank binders, add K_D Ranking: a concentration series on the same microarray gives an apparent K_D for every peptide.

The K_D is fitted from binding on a surface, with the peptide tethered and the protein in solution, and read after the detection reagent is added. Absolute values can differ from solution-phase measurements, in particular for fast-dissociating binders. The ranking is what the assay is built for. Confirm the top candidates in solution.

In projects with solution-phase reference values, the apparent K_D of the control peptides fell within the customer's SPR window. Rank differences within overlapping confidence intervals are not resolved. Use the array to decide which candidates deserve an SPR run, not to replace it.

Six concentrations, one per well, set per project and typically spanning several orders of magnitude, from the low nanomolar to the micromolar range. Binders that do not saturate within the series are reported as bounds, not as values.

The fit reports a Hill coefficient for every peptide. Values far from 1 flag avidity or non-specific binding, and those peptides are classed separately instead of being ranked as if they were 1:1 binders.

The peptide binds, but the series never reached its midpoint. The true K_D is at least the reported bound. A higher top concentration in the next series would measure it.

It depends on the protein more than on the array. The highest concentration must lie well above the K_D to saturate the surface-bound peptides, so the limit is set by how soluble the protein is and how much of it you have. Binders weaker than the series allows are reported as lower bounds.

Every peptide carries a reporter at its free end. When a protease cuts the peptide, the reporter is released and the spot loses its signal. One well receives assay buffer only and is the null: each treated well is measured against it, and reference spots inside every well correct for well-to-well differences. One scan after the exposure is enough.

Both are available, and the wells decide which. Vary the contact time and the fit gives a cleavage rate and a half-life per peptide: the question is how long a peptide survives. Vary the protease concentration at one contact time and the fit gives an apparent EC50, the concentration that cleaves half of it: the question is how little enzyme it takes. Both are rankings on a surface, not solution constants.

By comparing the sequences of cleaved and uncleaved peptides position by position, as a sequence logo. The assay ranks substrates and derives the motifs from the ranking. It does not localize the cut site within a peptide.

Each position shows the residues that favor cleavage above the axis and those that hinder it below. The height of a letter is its importance at that position.

No, and neither is the apparent EC50. The peptide is tethered to the surface, so both are rankings under the chosen condition, not solution constants. Overlapping confidence intervals are not ordered. Confirm the leads in solution, for example by LC-MS.

They fit different stages. LC-MS follows the intact peptide and its fragments in solution, so it gives a solution half-life and the cut site. It is the better choice for the last few candidates. The array compares up to 15,000 peptides on one microarray, all in the same plasma or protease, and ranks them with the motifs behind the ranking. On the array the C-terminus is anchored, so carboxypeptidases cannot reach it; in solution they can. The array narrows the library; LC-MS confirms the leads.

Not necessarily. Fouling by the sample, or enzymes that strip the reporter, also lower the signal. A fluorescent label also changes brightness with the peptide's charge and the sample matrix. A heat-inactivated sample run alongside separates proteolysis from such losses. Without it, the result is each peptide's susceptibility to signal loss in that matrix, and differences between related sequences point to the enzyme.

That the peptide lost no measurable signal over the tested window. It is a bound, not proof: the half-life is longer than the window. A longer time series would measure it.

Some antibodies bind an extended surface that no peptide reproduces, and the map stays empty for them. The result is then a validated negative, with the control peptides bright, and we say so. It is worth knowing before a crystallography campaign.

Yes. Sera and pooled samples are run like antibodies, one per well, and the map lists every candidate the sample recognizes across the whole antigen panel. Individual donors need individual wells.

As reliable as the model at that spot. The map flags epitopes in low-confidence regions, fragments farther apart than the structure allows, and residues buried in the fold. These checks are advisory: distances from a model are not experimental contacts.

Included. Processed results are workbooks with one row per peptide, plus the QC figures. The report explains the design, the quality control and the results, and a data dictionary defines every column.

Yes. The scanner exports of every spot are part of the delivery, with the peptide library and the spot-level table, so the analysis can be redone or extended.

Not from another project: reports fall under confidentiality agreements. Each service page shows an excerpt of a results table with the real column names. The report itself covers design, protocol, quality control, the results with figures, and a data dictionary.

Yes, and larger libraries require it. Up to 15,000 peptides are titrated directly. Above that, the library is screened by Binder Identification first, and the binders are titrated on a second microarray.

Where the library comes from, and what the result says. Binder Identification screens the library you send and calls each peptide bound or not. Epitope Mapping derives the candidates from the target protein's sequence and structure, and maps the hits back onto that protein.

They answer different questions. Phage display, like mRNA display, selects from a random library far larger than any array and returns the sequences enriched over the selection rounds. It is the better choice when there is no starting sequence and you need a binder from scratch. A peptide microarray tests a library you define, up to 200,000 peptides per array, and measures every peptide, binders and non-binders alike. It fits when the candidates already exist: designed by a model, variants of a lead, or tiles of a protein sequence. Many programs use both: display to find a hit, the array to map and optimize it.

• 1 well: 200,000 peptides
• 2 wells: 70,000 peptides each
• 6 wells: 15,000 peptides each
• 12 wells: 5,000 peptides each
• 18 wells: 3,200 peptides each

Up to 500,000 peptides per project from 14 days, on three microarrays in parallel; larger projects on request. One microarray holds from one condition with 200,000 peptides to 18 conditions with 3,200 peptides each: proteins, concentrations, buffers or controls.

One per well, up to 18 on one microarray, each seeing the identical candidate library, plus a buffer well as reference. Larger panels span several microarrays.

Your choice. Each well can carry its own library or a copy of another well's. Copies get the identical allocation of spots. Separate microarrays with identical content are possible too.

Up to 15,000 on one microarray. A titration needs every peptide in each of six wells, one protein concentration per well, so the library is the size of one well. The 200,000 peptides of the one-well layout meet a single concentration: one signal per peptide, a screen, not a titration.

Yes. Six is the minimum, and it keeps the library large, up to 15,000 peptides. For a finer ranking, the series can run over up to 18 wells, on fewer peptides per well.

At least two, ideally three, per peptide and condition. Where candidates are ranked by priority, the count can be tiered: more copies for the likeliest candidates, and single spots for the rest when capacity is short.

Yes: one list in duplicate, another in single spots, a control at its own count. New candidates added to an earlier list usually get a somewhat lower count than the original list.

Yes, as many as you like, as part of your library. They are placed at random among the test peptides. For human serum or plasma, we add positive controls that nearly every donor recognizes.

To keep position effects out of the results. A gradient across the slide spreads over all peptides instead of hitting one group. The allocation is reproducible: the same library gets the same positions, and peptides added later keep theirs. Layouts in a predefined order are not offered.

Yes. Send the protein sequences, or name the proteins, and we design the peptides to your plan. That can be overlapping tiles along a sequence, or substitution, combinatorial, random, truncation or deletion variants. Or send your own list, and the software does the rest.

A one-residue shift gives the highest resolution and is the default. A wider step shrinks the library, usually without losing epitope coverage, when more proteins must fit on the same microarray.

Each residue of a peptide is replaced by alanine in turn, one variant per position, so a 15-residue peptide adds 15 variants. The binding of the variants shows which residues carry the interaction.

Yes, up to one, within the project's capacity. The proteins are sliced into peptides, and each peptide is annotated with every protein it occurs in. A binding peptide then points to all the proteins that carry its sequence, which turns the screen into an off-target search. Filters keep a large proteome within capacity.

• Keep unique sequences only; each is annotated with every protein it belongs to.
• Keep one domain type: extracellular, transmembrane or intracellular.
• Keep surface-exposed stretches.
• Keep only the best-supported discontinuous candidates.
• Widen the tiling step.

For a large library and many samples, in two rounds. First, one large library is incubated with a pool of the samples, which separates non-binders from any binders. Then only the hits go onto a multi-well microarray, and each sample is incubated on its own. Every sample is characterized individually, at the cost of one more round of synthesis.

Yes, if the library is built from the structure. A linear scan tiles the sequence, so stretches that lie far apart in the chain never meet on one peptide. Candidates derived from the structure join them, and a hit names the fragments the antibody needs, drawn on the protein. An epitope that depends on the exact fold is better mapped by HDX-MS or a co-crystal structure.

Both, on one microarray. The overlapping scan finds continuous epitopes. The structure-derived candidates find the discontinuous epitopes that a scan overlooks: two fragments or a surface patch joined in one peptide, and cyclized to hold the shape.

An experimental structure from the Protein Data Bank when there is one, otherwise an AlphaFold model. You can also restrict the mapping to a domain or an extracellular region.

Two stretches that lie close in the structure are joined into one peptide, in the orientations their ends allow, and cyclized to hold the shape. A hit names both fragments by their residue positions.

A peptide assembled from residues that lie next to each other on the protein surface. It captures epitopes that span a β-sheet or several helices, beyond what two fragments reach.

Because it rules out more candidates. Only those the structure supports are kept, and a complete structure leaves fewer in doubt. Disordered regions have no coordinates, so continuous candidates cover them instead.

When you are certain the antibody binds a linear stretch, or already know the region and want the exact motif. A substitution or alanine scan can then go into the first round. Otherwise keep them: the microarray has room for them from the start, which spares a second round.

No. Overlapping continuous peptides present an epitope across a band of neighbors, and the joint of a two-fragment candidate faces the antibody. The caveat: a cyclic 15-mer need not reproduce the native fold, so an antibody that needs the exact 3D shape may not bind it.

One peptide per line, in a spreadsheet or plain text, in one-letter code. Mark per peptide what differs from the default: a free N-terminus, cyclization, or a custom building block by its code. Submit by email.

With its code in square brackets, inside the one-letter sequence: for example [Cit] for citrulline, [Nle] for norleucine, [dA] for D-alanine. An acetylated N-terminus is written [Ac]. We send the list of available codes, and new ones are added on request.

Your own names are kept. A peptide we derive from a protein is named by its type (continuous or discontinuous), the structure it came from, the protein and its residue positions. Every hit points back to its place in the sequence. A sequence found in several proteins carries all their names.

About 1 g per synthesis as a guide. The exact amount follows from the library: how many positions use the block, and its molecular weight. We estimate it when the library is submitted. You can send the block, or we source it at cost.

With the project submission: the final library and the order. The day plan then holds as long as its conditions are met: the purchase order is issued, the target protein has arrived (and is labeled, if we label it), and any custom building blocks are in the lab. Synthesis runs Day 1–7, the assay Day 8–10 and the analysis Day 11–13. The results are released on Day 14.

Yes. Design the library as for any project. We synthesize the arrays, check them and ship them, and you run the assay in your own lab. For now, arrays-only orders go through "Send a message". Recommendations for running the assay are shared on request; most labs use their own protocol.

The side on which the printed label reads the right way round.

In their sealed tubes, under inert atmosphere, at 4 to 8 °C, for up to six months from delivery. Transport at room temperature is fine.

From the layout file (GAL) that ships with the arrays. GAL-compatible software imports it and locates each block. Control spots mark the frame of every block, and the top-left corner spot fixes the orientation.

• a blocking buffer
• PBS with 0.05 % Tween 20 for washing
• an incubation tray that fits the layout
• your detection reagents, for example labeled streptavidin, or a labeled anti-HA antibody to read the controls
• a microarray centrifuge for drying
• a fluorescence scanner (step 5.4)

We share the products we use on request.

Not all of them. The tray has more wells than the array has fields, and its outer rows fall outside the peptide area. Which rows to skip depends on the layout; we tell you for yours.

Just before the scan, evenly and without stains, in a microarray centrifuge or a stream of inert gas. Our recommendations are shared on request.

As short as the reagent allows. Everything between the end of the incubation and the scan gives weak binders time to dissociate: the washes, the stain and the washes after it. Our recommendations are shared on request.

A fluorescence microarray scanner with a red channel around 635 nm and a resolution of 5 µm per pixel or finer. We recommend the InnoScan 910 or 1100 (Innopsys), which have low-noise detectors and are common second-hand. A GenePix (Molecular Devices) works if it reaches the resolution.

No. A spot is about 40 × 40 µm, so at 10 µm per pixel it covers only a few pixels and cannot be resolved. For an end-point readout, you can scan at 10 µm now and rescan the dried slide later at 5 µm or finer.

Any microarray image software that imports a GAL layout file. It extracts the signal of every spot. The analysis of chapter 3 can then run on our side (step 5.6).

• Ask a scanner maker (Innopsys, Molecular Devices) for access. They can usually point you to a nearby instrument.
• Run the assay yourself and send the arrays back to us for scanning. This is a one-shot assay, with little room to re-incubate.
• Isolate the antibodies from the samples with Protein A/G beads and send them, and we run the assay.

Back in its tube, sealed under inert atmosphere at 4 to 8 °C, for up to six months from delivery. After human samples this matters most: residual proteases can cleave some peptides.

In principle, yes, after the bound protein is stripped off. Whether it works depends on the analyte and its affinity: stripping removes much, but rarely all, of the bound antibody. Reuse also lowers the sensitivity of the next run. A control scan must confirm the old signal is gone before re-probing. A fresh array remains the more reliable choice.

Yes. Send the scanner exports with the layout file, your peptide list and the conditions of every well: the protein, and its concentration or contact time. The analysis is the same as for the assays we run (chapter 3). You receive the processed results and a report.