Overlapping peptide generator

How to generate overlapping peptides

Paste a protein sequence or upload a FASTA file to generate an overlapping peptide library. Set Peptide length and Overlap, then select Generate peptides. The result opens on FASTA, with sequences and coordinates in Peptides. Download FASTA or CSV for the library and JSON for the complete design record. Generation runs locally in your browser.

The first 54 residues of the human insulin precursor, UniProt P01308, produce six peptides with the default length of 18 and overlap of 10:

>P01308 insulin precursor residues 1-54
MALWMRLLPLLALLALWGPDPAAAFVNQHLCGSHLVEALYLVCGERGFFYTPKT

The FASTA download contains:

>peptide_1 block=1 reference=1-18 sources=1
MALWMRLLPLLALLALWG
>peptide_2 block=2 reference=9-26 sources=1
PLLALLALWGPDPAAAFV
>peptide_3 block=3 reference=17-34 sources=1
WGPDPAAAFVNQHLCGSH
>peptide_4 block=4 reference=25-42 sources=1
FVNQHLCGSHLVEALYLV
>peptide_5 block=5 reference=33-50 sources=1
SHLVEALYLVCGERGFFY
>peptide_6 block=6 reference=41-54 sources=1
LVCGERGFFYTPKT

The last fragment has 14 residues. All coordinates are one-based and inclusive. sources=1 identifies the first input record; the table and JSON preserve its full name.

Multiple proteins in one FASTA file

Keep Aligned sequences off to tile each record independently. These human insulin B and A chains produce five peptides with the same default settings:

>P01308 insulin B chain
FVNQHLCGSHLVEALYLVCGERGFFYTPKT
>P01308 insulin A chain
GIVEQCCTSICSLYQLENYCN

Coordinates restart at 1 for each submitted record; peptide IDs continue across the run. Use FASTA headers to separate proteins. With alignment mode off, raw lines without headers are joined into one sequence.

Input

Whitespace and lowercase letters are normalized. Empty records, records containing only gaps, stop markers, numbering and unsupported punctuation cause an error. Gaps require alignment mode. Convert GenBank, Clustal, Stockholm and other formatted files to FASTA first. The generator does not align sequences; use a protein alignment when needed.

Translate a coding DNA sequence with DNA to protein, or extract a protein chain from a structure with PDB to FASTA. For deliberate removal or replacement of unwanted characters, use Filter protein and review its changes before generating the library.

Settings

Open Terminal restrictions for terminal residues, length adjustments and the proline rule. Output options contains hydropathy, duplicates and gaps. All lengths are whole numbers of amino acids.

Enter forbidden residues as one-letter codes, either together (GPEDQNTSC) or separated by spaces (G P E D Q N T S C). Lowercase is accepted and repeated letters are ignored. Do not use commas or three-letter residue names. Nonstandard residue codes are allowed in input sequences but rejected in these settings.

Results and downloads

The tabs are FASTA, Peptides, Files and Logs, in that order. FASTA opens first with a copyable preview. The Peptides tab shows up to 5,000 rows; downloaded peptide files contain every row.

The Files tab contains these downloads. The filename prefix comes from the uploaded file, or is peptides for pasted input.

Read run.log in Logs and download it there or from Files. When the generator rejects input, settings or a design exceeding its limits, the log gives a recovery suggestion and no partial peptide files are returned. Checks that prevent generation from starting do not create output files.

CSV column headers use field names such as peptideId, referenceStart, sourcePositions and duplicateOf; the table displays the readable labels above. JSON represents unavailable hydropathy as null, while CSV leaves the cell empty.

How peptide boundaries are selected

Peptide tiling divides a protein into windows that share part of their sequence. With length 18 and overlap 10, fixed windows start every 8 residues: positions 1, 9, 17 and so on. For an ungapped protein with terminal adjustments off, this covers every contiguous 11-residue stretch, which is useful when planning an epitope-mapping library.

The method follows the boundary-selection procedure described in Los Alamos PeptGen. It tries the requested length, then permitted shorter lengths, then longer ones to avoid forbidden C-terminal residues. If all fail, it retains a flagged peptide, optionally applying the proline fallback. For one-based coordinates, the next start is the previous end minus overlap plus one; an N-terminal restriction can move that start left.

The remaining tail is emitted when it is shorter than length plus maximum extension. It can violate terminal restrictions and is flagged accordingly. Like PeptGen, an exact-length final full peptide can be followed by a fragment containing only the overlap. Review short fragments before ordering a synthesis library.

The final fragment can also be longer than the target. For the 30-residue insulin B chain above, length 18, overlap 10 and Lengthen by set to 5 produce peptides at positions 1-18 and 9-30, with lengths of 18 and 22. The remaining 22 residues are below the final-fragment threshold of 23, so they are retained together even with no terminal restrictions.

If restrictions prevent progress, the run stops with an error rather than skipping a region. Reduce overlap or relax the restrictions.

Alignment behavior and PeptGen differences

The first sequence determines the boundaries. Every other sequence contributes the same alignment columns, so insertions and deletions change variant lengths. Gaps before a reference residue belong to the window containing that residue; this retains insertions between windows even at zero overlap. Leading and trailing insertions are retained as well. Terminal restrictions apply to the reference only.

This independent implementation differs from PeptGen's simple gap-removal output: variants are not extended or split to force the reference length. Entirely gapped slices are omitted from peptide files and recorded in JSON. Duplicate sequences at different blocks remain separate.

For example, this insulin B-chain fragment is aligned to a synthetic variant with one inserted alanine:

>P01308 B-chain fragment
FVNQHL-CGSHLVE
>synthetic insertion
FVNQHLACGSHLVE

Enable Aligned sequences, set Peptide length to 8 and Overlap to 3, and leave the other settings at their defaults. The run produces six peptides. Its first block illustrates the difference between reference positions and alignment columns:

The inserted alanine remains in the variant peptide. The final block contains LVE from both records. Choosing Remove within each block keeps one LVE row with two source occurrences, reducing the output to five peptides.

Interpreting hydropathy

Hydropathy is the arithmetic mean of the standard Kyte-Doolittle residue values. Higher values indicate a more hydrophobic composition. It is dimensionless, and does not predict solubility or antigenicity. CSV and JSON retain numeric precision; the text report displays two decimal places.

For example, ACDEFGHI has a mean of 0.125. A peptide containing X or another nonstandard residue has no reported value. A hydropathy plot shows how this property varies across the parent protein.

Which peptide tool should I use?

FAQ

How do I generate 15-mers overlapping by 11 residues?

Set Peptide length to 15 and Overlap to 11. Leave terminal restrictions empty and shortening and extension at zero for fixed windows. Starts advance by four residues, followed by a final fragment when needed.

Why is the last peptide shorter than the requested length?

The generator retains the remaining tail. With the default settings, an input of exactly 18 residues produces the full 18-mer and a final 10-residue overlap fragment. Review Length (aa) and Notes to decide whether to include short fragments in the synthesis order.

Why does a peptide still end in a forbidden residue?

Terminal restrictions are preferences within the allowed length range. If no permitted length has an allowed ending, the peptide is retained with a warning; the final fragment is also retained even if it violates the restrictions. Allow shortening or extension and inspect Notes before ordering.

Can I remove all duplicate peptides across the library?

Duplicate removal applies only within each alignment block. Identical sequences from different blocks or independent protein records remain separate so their positions are preserved.

Does this predict epitopes?

No. It creates overlapping peptide sets for experimental planning. It does not predict MHC binding, immunogenicity, or whether a peptide can be synthesized successfully. DeepImmuno scores peptide-HLA immunogenicity for 9- or 10-residue peptides; select peptides of those lengths and specify an HLA allele for that analysis.

Can I use several unrelated proteins?

Yes. Submit multi-record FASTA with Aligned sequences off. Each protein is tiled independently, and every peptide retains its source name and coordinates.