# Structure-based sequence alignment Use three-dimensional geometry to derive residue correspondence when sequence similarity alone is too weak or ambiguous. [Open workflow](https://proteiniq.io/app/workflows/templates?template=structure-based-sequence-alignment)[Contact sales](https://proteiniq.io/contact) ## More sequence alignment workflows ## Tools for structure-based sequence alignment Use these methods to prepare inputs, run the core analysis, inspect outputs, and validate the evidence described in this workflow. [![USAlign](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fusalign.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### USAlign Align macromolecular structures and derive residue correspondence](https://proteiniq.io/app/usalign)[![FoldSeek](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Ffoldseek.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### FoldSeek Find and compare structurally similar proteins](https://proteiniq.io/app/foldseek)[![PDB to FASTA converter](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fpdb-to-fasta.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### PDB to FASTA converter Extract sequences from structures before comparison](https://proteiniq.io/app/pdb-to-fasta)[![MAFFT](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fmafft.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### MAFFT Create configurable protein, DNA, or RNA alignments](https://proteiniq.io/app/mafft)[![Clustal Omega](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fclustal-omega.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### Clustal Omega Create scalable protein or nucleotide multiple-sequence alignments](https://proteiniq.io/app/clustal-omega)[![MUSCLE5](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fmuscle5.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### MUSCLE5 Generate conventional or ensemble multiple-sequence alignments](https://proteiniq.io/app/muscle5)[![HMMER](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fhmmer.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### HMMER Find homologs with profile hidden Markov models](https://proteiniq.io/app/hmmer)[![MMseqs2](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fmmseqs2.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### MMseqs2 Search and cluster large protein or nucleotide sequence sets](https://proteiniq.io/app/mmseqs2)[![StringZilla v5](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fstringzilla.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### StringZilla v5 Calculate pairwise global, local, or edit-distance score matrices](https://proteiniq.io/app/stringzilla)[![FastTree](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Ffasttree.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### FastTree Estimate trees from large sequence alignments](https://proteiniq.io/app/fasttree)[![IQ-TREE](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fiq-tree.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### IQ-TREE Infer maximum-likelihood phylogenies from alignments](https://proteiniq.io/app/iq-tree)[![RAxML-NG](https://proteiniq.io/_next/image?url=%2Fimages%2Ftools%2Fraxml-ng.webp&w=96&q=75&dpl=dpl_J8tu6aF3F41kEvMFgYtuMXes1ynJ) ### RAxML-NG Run maximum-likelihood phylogenetic analysis](https://proteiniq.io/app/raxml-ng) ## What is structure-based sequence alignment? Structure-based sequence alignment is the process of arranging protein sequences by matching residues that occupy corresponding positions in three-dimensional structures. It is especially useful for remote homologs whose folds are conserved despite low sequence identity. In ProteinIQ, USAlign performs structural superposition and returns the resulting residue alignment, TM-scores, RMSD, sequence identity, and superposed coordinates. Choose structure-based alignment when compatible experimental or predicted structures exist and sequence-only methods disagree or fail to recover a conserved fold. The method can clarify core correspondence across remote homologs, circular permutations, oligomers, and nucleic-acid structures, depending on the selected USAlign mode. Structural alignment is not independent of structure quality. Missing residues, alternate conformations, domain movements, chain selection, prediction uncertainty, and different biological assemblies can all change the correspondence. Review coverage, both normalized TM-scores, RMSD, aligned length, sequence identity, and the superposed coordinates together. ### When to use structure-based sequence alignment Use this approach when the comparison scope is clear: - Two structures are available despite weak sequence identity. - Residue mapping depends on a conserved fold. - A structural superposition will guide annotation review. ### Strengths and limitations A three-dimensional superposition can locate corresponding core residues after sequence divergence obscures them. US-align returns a residue mapping and score normalizations that can be inspected alongside the actual aligned coordinates. A shared fold does not mean every loop or active-site residue corresponds. Flexible domains, different oligomeric states and incomplete models can distort a rigid superposition. [TM-score](https://proteiniq.io/guides/tm-score-vs-rmsd) and RMSD emphasize different aspects of the aligned region. ## Structure-based sequence alignment methods USAlign searches for a residue correspondence and rigid-body superposition that optimize a length-normalized TM-score. TM-score and RMSD summarize different aspects of the result and should not be treated as interchangeable. [FoldSeek](https://proteiniq.io/app/foldseek) can identify structurally similar candidates at scale, while USAlign is suited to detailed pairwise superposition. Sequence extraction may help compare structure-derived and sequence-only alignments without conflating them. ## Structure-based sequence alignment applications A structure-derived mapping can connect residues across remote homolog candidates or compare models with an experimental reference. It is particularly useful for a conserved core when ordinary sequence alignment places many gaps ambiguously. Functional annotation transfer should use local geometry and chemistry around the proposed site. A global fold match can coexist with different ligands, mechanisms or evolutionary histories. ## How to run structure-based sequence alignment online The featured workflow gives two PDB structures to US-align and retains the superposition and residue correspondence. 1. **Choose structures.** Select the relevant chains, constructs and biological states. 2. **Check coordinates.** Note missing residues, low-confidence segments and major domain movement. 3. **Run pairwise superposition.** Use [US-align](https://proteiniq.io/app/usalign) on mobile and reference PDB inputs. 4. **Inspect mappings.** Read aligned length, both TM-score normalizations, RMSD and gaps with the superposed coordinates. 5. **Review local sites.** Check whether catalytic or binding residues truly correspond before transferring labels. 6. **Export evidence.** Keep structures, chain choices, alignment and normalization settings together. ## How to interpret structure-based sequence alignment results Inspect aligned length and both TM-score normalizations because unequal chain lengths can produce asymmetric values. RMSD should always be read with the number and fraction of aligned residues. Do not transfer catalytic, binding, or numbering annotations across a gap or poorly superposed loop without local evidence. Conserved global folds can support very different biochemical roles. ## How structure-based sequence alignment works USAlign superposes mobile and reference structures and derives a residue alignment from their three-dimensional correspondence. 1. **Choose structures.** Choose structures representing the intended state, construct, and biological assembly. 2. **Select chains.** Set chains, molecule type, oligomer handling, and any circular-permutation mode. 3. **Run USAlign.** Run USAlign and preserve both score normalizations and aligned coordinates. 4. **Inspect superposition.** Inspect coverage, RMSD, gaps, flexible regions, and the superposed model. 5. **Transfer cautiously.** Transfer residue annotations only across well-supported structural correspondence. ## Inputs and outputs Check formats before running, then inspect and download the result from every workflow step. ### Inputs #### Alignment input `FASTA``PDB``mmCIF` Two protein, RNA, or DNA structures in PDB or mmCIF format. ### Outputs #### Alignment outputs `FASTA``CSV``TSV``PDB``JSON` Structure-derived residue alignment, TM-scores, RMSD, aligned length, identity, and superposed coordinates. ## Frequently asked questions AI answers from our docs. No sign-in needed. ## Run your first analysis today Start with 200 free credits and no installs. Upgrade when your project needs more compute. [Start for free](https://proteiniq.io/sign-up)[View pricing](https://proteiniq.io/pricing)