# How to Use DALI for Protein Structure Comparison: A Practical Tutorial for Detecting Structural Similarity

Researchers who need to compare a query protein structure against the Protein Data Bank to find structural neighbors face a specific problem: DALI is a powerful tool, but its output can be difficult to interpret without a clear workflow. This tutorial provides a step-by-step approach to running DALI, preparing input files, interpreting Z-scores and alignments, and visualizing results, with attention to common pitfalls and practical decision criteria.

DALI (Distance-matrix ALIgnment) is a structural comparison method that identifies proteins with similar three-dimensional folds even when their amino acid sequences have diverged substantially. This capability matters because protein functional domains can maintain conserved 3D structures throughout evolution even when primary sequences differ considerably, a principle that has been demonstrated in viral protein classification where structural homology reveals deep taxonomic relationships that sequence-based methods miss [8]. For biology students, researchers, and laboratory professionals working in structural bioinformatics, understanding how to run DALI correctly and interpret its output is essential for tasks ranging from functional annotation to evolutionary analysis.

## At a Glance

The table below summarizes the key decisions and expected outcomes when using DALI for protein structure comparison.

| Workflow Stage | Primary Decision | Expected Output | Common Error |
| --- | --- | --- | --- |
| Input preparation | Select PDB ID or upload coordinate file | Clean structure file ready for submission | Missing chain identifiers or heteroatoms causing rejection |
| Server submission | Choose pairwise or search mode | Job ID and queue position | Submitting without specifying chain selection |
| Z-score interpretation | Set significance threshold | List of structural neighbors ranked by Z-score | Treating Z-scores below 2 as meaningful |
| Alignment analysis | Review residue-level mapping | Superposed coordinates and alignment statistics | Ignoring alignment coverage when assessing similarity |
| Visualization | Select viewing tool | Structural superposition image | Misinterpreting RMSD without considering alignment length |

## Understanding DALI and Its Role in Structural Biology

DALI operates on the principle that protein structures can be compared by examining their internal distance matrices. The method calculates a similarity score based on the equivalence of residue-residue distances between two proteins, allowing it to detect structural similarities that are invisible to sequence alignment tools. This approach is particularly valuable for studying proteins where sequence conservation is low but structural conservation is high.

The structural biology community has developed multiple tools for structure comparison, and DALI remains one of the most widely used because of its sensitivity and the availability of a public server. The DALI server accepts a query structure and searches it against the entire Protein Data Bank, returning a ranked list of structural neighbors. Each match includes a Z-score, which measures the statistical significance of the structural similarity, and an alignment that shows which residues correspond between the two structures.

For researchers working with protein structure prediction, DALI serves as a validation tool. Predicted structures can be compared against experimentally determined structures or against the broader PDB to assess whether the predicted fold matches known structural families. In the context of molecular docking interpretation, DALI can help identify whether a protein of interest shares structural features with proteins of known function, providing clues about binding sites and interaction interfaces.

The integration of structural comparison tools into larger workflows has become increasingly common. For example, the Scipion workflow engine now includes plugins that enable researchers to combine structural dynamics analysis with cryo-electron microscopy image processing, demonstrating how structural bioinformatics tools are being connected into comprehensive analysis pipelines [11]. DALI fits into this ecosystem as a foundational comparison method that can feed into downstream analyses.

## Preparing Your Input Structure

Before submitting a query to the DALI server, you must ensure that your input structure is properly prepared. The quality of your input directly affects the quality of your results, and several common preparation steps can prevent failed submissions or misleading outputs.

### Selecting the Source of Your Structure

Your query structure can come from several sources. If you are working with a structure already deposited in the Protein Data Bank, you can simply provide the four-character PDB ID. The National Center for Biotechnology Information provides access to the Molecular Modeling Database and related structure resources that can help you locate and retrieve structures of interest [1]. If you have determined a new structure or generated a model, you will need to upload a coordinate file in PDB or mmCIF format.

For researchers working with predicted structures, it is important to recognize that DALI was designed primarily for experimentally determined structures. Predicted models can be submitted, but the results should be interpreted with additional caution because prediction errors can create artificial structural similarities or obscure genuine ones.

### Cleaning the Coordinate File

A raw coordinate file often contains more information than DALI needs. Water molecules, ligands, and other heteroatoms can interfere with the comparison, particularly if they occupy positions that create spurious distance relationships. Most structural biology software packages provide tools for removing these components, and the process is straightforward.

Chain selection is another critical step. Many structures in the PDB contain multiple chains, and comparing a multimeric complex against the entire PDB can produce confusing results. For most applications, you should select a single chain for the DALI search. This is particularly important when your query is a domain or subunit that may have structural homologs in different oligomeric contexts.

The European Bioinformatics Institute offers training materials on structural bioinformatics that cover structure preparation and quality assessment [2]. These resources can help you understand what constitutes a usable structure file and how to identify potential problems before submission.

### Format Considerations

DALI accepts standard PDB format and mmCIF format files. If you are uploading a file, ensure that it contains complete coordinate information for all residues you intend to include in the comparison. Truncated files or files with missing residues can lead to incomplete alignments that underestimate structural similarity.

For structures determined by cryo-electron microscopy, you may need to consider whether the deposited coordinates include all residues or only those that were resolved in the density map. Disordered regions that lack coordinates will be excluded from the comparison, which is appropriate but should be noted when interpreting results.

## Running DALI on the Public Server

The DALI server provides a web interface that accepts query structures and returns results, typically within minutes to hours depending on server load. Understanding the submission process and the options available helps you design comparisons that answer your specific research question.

### Submission Modes

DALI offers two primary modes of comparison. The first is a search mode, where your query structure is compared against all structures in the PDB. This mode is useful for identifying structural neighbors and discovering proteins that may share functional characteristics with your query. The second is a pairwise mode, where you specify two structures to compare directly. This mode is useful for testing a specific hypothesis about the relationship between two proteins.

For most discovery-oriented applications, the search mode is the appropriate starting point. The server will return a list of the top structural matches, ranked by Z-score, along with alignment details for each match. For hypothesis testing, pairwise mode provides a focused comparison that can be completed more quickly.

### Job Submission and Monitoring

When you submit a job to the DALI server, you will receive a job identifier that you can use to retrieve your results. The server processes jobs in a queue, and wait times depend on the number of pending submissions and the complexity of your query structure. Large structures or structures with many chains may take longer to process.

The Galaxy Training Network provides guidance on running bioinformatics analyses in reproducible workflows, and the principles of job tracking and result retrieval apply to DALI as well [4]. Keeping a record of your job identifiers and submission parameters is good practice, particularly if you need to reproduce your analysis or share your methods with collaborators.

### Parameter Selection

The DALI server offers several parameters that control the sensitivity and scope of the search. The most important of these is the minimum Z-score threshold, which determines which matches are reported. Lowering the threshold increases sensitivity but also increases the number of false positives. Raising the threshold produces a more conservative list of matches but may miss biologically meaningful similarities.

Another parameter controls whether the search considers all chains in the query structure or only a specified chain. For most applications, specifying a single chain is appropriate. If you are studying a protein complex, you may want to search with the full complex to identify other complexes with similar quaternary structures, but this approach requires careful interpretation because the comparison will reflect both subunit structure and assembly geometry.

## Interpreting DALI Output

The output from a DALI search contains several components that must be interpreted together to draw meaningful conclusions about structural similarity. The most important of these are the Z-score, the alignment coverage, and the root-mean-square deviation (RMSD) of the aligned residues.

### Understanding Z-Scores

The Z-score is the primary measure of statistical significance in DALI results. It represents how many standard deviations the observed similarity score is above the mean score expected for unrelated proteins of similar size. Higher Z-scores indicate more significant structural similarity.

A common threshold for considering a match significant is a Z-score of 2.0, but this threshold is not absolute. The significance of a given Z-score depends on the size of the query structure and the size of the database being searched. For a typical protein domain of 100 to 300 residues, Z-scores above 2.0 generally indicate genuine structural similarity, while scores between 2.0 and 3.0 may require additional evidence to confirm biological relevance.

The relationship between Z-score and structural similarity is not linear. Two proteins can have the same Z-score but different alignment lengths and RMSD values, and these additional metrics must be considered when assessing the quality of a match. A high Z-score with a short alignment may indicate that only a portion of the structures are similar, while a moderate Z-score with a long alignment may indicate more extensive structural conservation.

### Alignment Coverage and RMSD

The alignment produced by DALI shows which residues in the query structure correspond to which residues in the target structure. The coverage of the alignment, expressed as the fraction of query residues that are aligned, is an important quality metric. A high Z-score with low coverage suggests that only a small region of the query matches the target, which may still be biologically meaningful if that region corresponds to a functional domain.

RMSD measures the average distance between equivalent atoms after optimal superposition. Lower RMSD values indicate better structural agreement. However, RMSD must be interpreted in the context of alignment length. A low RMSD over a short alignment is less informative than a moderate RMSD over a long alignment, because the latter indicates that the overall fold is conserved.

The structural biology community has developed additional metrics for assessing structural similarity, including the TM-score, which is used in tools like MPACT for viral classification [8]. TM-score normalizes for protein size and provides a more interpretable measure of topological similarity than RMSD alone. While DALI does not directly report TM-scores, understanding this metric helps you place DALI results in the broader context of structural comparison methods.

### Visualizing Alignments

DALI results include coordinate files for the superposed structures, which can be visualized in molecular graphics programs. Visualization is essential for understanding the spatial relationship between the query and target structures, particularly when the alignment reveals that similar regions are distributed across the protein instead of contiguous.

When visualizing DALI alignments, pay attention to the regions that are not aligned. These regions may represent insertions, deletions, or structural rearrangements that distinguish the two proteins. In some cases, unaligned regions correspond to functional differences, such as the presence of additional domains or loops that confer unique binding properties.

The Scipion-EM-ProDy plugin demonstrates how structural analysis tools can be integrated into graphical workflow environments, making it easier for researchers to combine DALI comparisons with dynamics analysis and other structural bioinformatics methods [11]. For researchers who need to perform structural comparisons repeatedly, investing time in learning such workflow tools can improve efficiency and reproducibility.

## Practical Workflow for Structural Neighbor Detection

A systematic workflow for using DALI to detect structural neighbors involves several stages, from initial query preparation to final interpretation. The following steps provide a practical framework that can be adapted to specific research questions.

### Step 1: Define Your Research Question

Before running DALI, clarify what you want to learn from the comparison. Are you trying to identify the structural family of a newly determined protein? Are you testing whether two proteins share a common ancestor? Are you looking for proteins with similar binding sites to guide mutagenesis experiments? The specific question determines which mode of DALI you should use and how you should interpret the results.

For functional annotation, a search against the entire PDB is appropriate. For evolutionary analysis, you may want to restrict your search to structures from specific organisms or taxonomic groups. For docking interpretation, you may want to compare your protein against known receptor-ligand complexes to identify structural features associated with binding.

### Step 2: Prepare and Validate Your Query

Ensure that your query structure is complete and correctly formatted. Check that all residues have coordinates and that the chain you intend to compare is properly identified. If your structure contains multiple conformations, such as those from NMR ensembles, select a representative model for the comparison.

Validation of your query structure is important, particularly for predicted models. Tools that assess model quality, such as those available through the Protein Data Bank and NCBI structure resources, can help you identify problematic regions that may affect DALI results [1]. If your model has low confidence in certain regions, consider whether those regions should be excluded from the comparison.

### Step 3: Submit the DALI Search

Submit your query to the DALI server, specifying the appropriate parameters for your research question. Record the job identifier and the submission parameters for your records. If you are running multiple queries, maintain a spreadsheet that tracks each job, its parameters, and the date of submission.

### Step 4: Retrieve and Organize Results

When your job completes, download the results, including the ranked list of matches, the alignment files, and the superposed coordinates. Organize these files in a directory structure that allows you to retrieve them easily for analysis and sharing with collaborators.

The nf-core documentation emphasizes the importance of reproducible workflows in bioinformatics, and the same principles apply to structural comparison analyses [5]. Documenting your DALI runs, including the exact parameters used and the version of the server, enables others to reproduce your results and allows you to update your analysis when new structures are added to the PDB.

### Step 5: Filter and Rank Matches

Apply your Z-score threshold to filter the list of matches. For most applications, focus on matches with Z-scores above 2.0, but examine matches with lower scores if they involve proteins with known functional relationships to your query. Consider the alignment coverage and RMSD for each match to assess the quality of the structural similarity.

### Step 6: Visualize and Interpret

Visualize the top matches to understand the structural relationships. Examine the aligned regions and the unaligned regions to identify conserved cores and variable loops. Consider whether the structural similarity corresponds to known functional similarities or suggests new functional hypotheses.

### Step 7: Document and Report

Record your findings in a format that supports your research conclusions. Include the DALI parameters, the top matches with their Z-scores and alignment statistics, and your interpretation of the structural relationships. This documentation is essential for publications, grant reports, and collaboration with other researchers.

## Options and Tradeoffs in Structural Comparison

DALI is one of several structural comparison tools available to researchers, and each tool has strengths and limitations that make it more or less suitable for specific applications. Understanding these tradeoffs helps you choose the right tool for your research question and interpret results appropriately.

### DALI Compared to Sequence-Based Methods

Sequence alignment tools such as BLAST and HMMER identify homologous proteins based on amino acid sequence similarity. These methods are fast and sensitive for closely related proteins but fail to detect homology when sequences have diverged beyond approximately 20 to 30 percent identity. Structural comparison methods like DALI can detect homology at much greater evolutionary distances because structure is more conserved than sequence.

The value of structural comparison for detecting distant relationships has been demonstrated in viral taxonomy, where proteins with highly divergent sequences can share conserved 3D structures that reveal deep evolutionary connections [8]. For researchers studying proteins with no clear sequence homologs, DALI provides a way to connect their protein to known structural families.

### DALI Compared to Other Structural Comparison Tools

Several other structural comparison tools are available, including TM-align, CE, and Foldseek. Each uses a different algorithm and produces results that may differ in ranking and alignment details. TM-align is particularly popular for pairwise comparisons and produces TM-scores that are normalized for protein size. Foldseek uses a different approach based on 3Di alphabet encoding and is extremely fast, making it suitable for searching very large databases.

The choice of tool depends on your specific needs. DALI is well suited for discovering structural neighbors in the PDB because its Z-score provides a statistical measure of significance. TM-align is well suited for pairwise comparisons where you want a normalized measure of topological similarity. Foldseek is well suited for very large-scale searches where speed is a priority.

For researchers who need to integrate multiple metrics, tools like MPACT combine sequence-based and structure-based measures to provide a more complete picture of protein relationships [8]. This integrated approach can be particularly valuable for classification tasks where no single metric is sufficient.

### Tradeoffs in Sensitivity and Specificity

Structural comparison tools face an inherent tradeoff between sensitivity and specificity. More sensitive methods detect more true positives but also produce more false positives. Less sensitive methods produce fewer false positives but may miss genuine similarities.

DALI's Z-score provides a principled way to manage this tradeoff by selecting a threshold that balances sensitivity and specificity for your application. For discovery-oriented searches, a lower threshold may be appropriate to avoid missing potential homologs. For hypothesis testing, a higher threshold provides more confidence that reported matches are genuine.

## Observations and Measurements in DALI Analysis

Systematic observation of DALI results across multiple queries reveals patterns that can guide interpretation and improve the reliability of structural comparison analyses.

### Z-Score Distributions

The distribution of Z-scores for a typical DALI search shows a clear separation between significant matches and background noise. Genuine structural homologs typically have Z-scores well above 2.0, while unrelated proteins cluster near zero. The gap between these populations provides a natural threshold for distinguishing meaningful matches from random similarities.

However, the Z-score distribution depends on the size and fold of the query structure. Small proteins and proteins with common folds tend to have higher background Z-scores because there are more proteins in the database with similar sizes and topologies. Large proteins and proteins with unusual folds tend to have lower background Z-scores, making it easier to identify genuine matches.

### Alignment Quality Metrics

The relationship between Z-score, alignment coverage, and RMSD provides insight into the nature of structural similarity. High Z-scores with high coverage and low RMSD indicate strong overall structural conservation. High Z-scores with low coverage may indicate that a specific domain or region is conserved while the rest of the protein differs. Moderate Z-scores with high coverage may indicate weak but extensive structural similarity.

Recording these metrics for each match in your DALI results allows you to compare matches systematically and identify those that are most likely to be biologically meaningful. A spreadsheet that tracks Z-score, coverage, RMSD, and alignment length for each match provides a useful basis for decision-making.

### Consistency Across Query Variations

Running DALI with slightly different query preparations can reveal the robustness of your results. For example, comparing a structure with and without a flexible loop region can show whether the loop contributes to or detracts from the detected similarity. Comparing different chains of a multimeric protein can reveal whether structural similarity is consistent across subunits.

These consistency checks are particularly important for predicted structures, where small conformational differences can affect DALI results. If a predicted structure produces different structural neighbors than the experimentally determined structure, the differences may indicate regions of the model that are unreliable.

## Records and Documentation for Reproducibility

Reproducibility is a core principle of bioinformatics research, and structural comparison analyses are no exception. Maintaining detailed records of your DALI analyses enables you to reproduce your results, share your methods with collaborators, and update your analyses as new structures become available.

### What to Record

For each DALI analysis, record the following information: the query structure identifier or file name, the chain used for comparison, the DALI server version and submission date, all parameters used for the search, the job identifier, and the date results were retrieved. Also record the version of the PDB that was searched, because the database changes over time and results from different dates may not be directly comparable.

The Carpentries lessons on reproducible research emphasize the importance of documentation and version control for computational analyses [6]. Applying these principles to structural comparison work means keeping your DALI inputs, parameters, and outputs in a version-controlled repository that tracks changes over time.

### Managing Results Over Time

The PDB grows continuously, and a DALI search performed today may produce different results than the same search performed six months from now. If you are using DALI results to support ongoing research, plan to rerun your searches periodically to ensure that your conclusions remain current.

For large-scale analyses involving many queries, consider using workflow management tools that automate the submission and retrieval of DALI jobs. The Galaxy Training Network provides guidance on building reproducible analysis workflows that can be applied to structural bioinformatics [4]. Similarly, the nf-core documentation describes standards for community pipelines that emphasize reproducibility and portability [5].

### Sharing Methods and Results

When publishing research that relies on DALI results, include sufficient detail about your methods to allow others to reproduce your analysis. Specify the DALI server URL, the version of the server if known, the query structure identifier, the chain used, and the parameters selected. If space permits, include the full list of significant matches in supplementary materials.

The Bioconductor project provides an example of how the bioinformatics community documents and shares analysis methods through standardized package documentation and vignettes [3]. While Bioconductor focuses on genomic analysis, the principles of clear documentation and reproducible workflows apply equally to structural bioinformatics.

## Common Failure Patterns and How to Avoid Them

Several recurring problems can compromise DALI analyses. Recognizing these failure patterns helps you avoid them and interpret results correctly when they do occur.

### Submitting Poorly Prepared Structures

The most common cause of failed DALI submissions is a malformed or incomplete coordinate file. Structures that contain only C-alpha atoms, structures with missing residues, and structures with incorrect chain identifiers can all cause errors. Before submitting, validate your structure file using tools that check for completeness and format correctness.

For structures derived from cryo-electron microscopy, be aware that deposited coordinates may not include all residues visible in the density map. The resolution of the map and the local quality of the density determine which residues can be modeled reliably. If your structure has regions with poor density, consider whether those regions should be included in the DALI comparison.

### Misinterpreting Z-Scores

Treating Z-scores as absolute measures of similarity instead of statistical measures of significance leads to misinterpretation. A Z-score of 2.0 for a small protein may be less meaningful than a Z-score of 2.0 for a large protein, because the background distribution of scores differs. Always consider Z-scores in the context of your query structure and the database being searched.

Another common error is ignoring the alignment coverage when assessing significance. A high Z-score with very low coverage may indicate that only a small fragment of the query matches the target, which may or may not be biologically meaningful. Examine the alignment to understand which regions are similar and whether those regions correspond to functional domains.

### Overlooking the Effects of Database Growth

The PDB grows continuously, and the results of a DALI search depend on the database version. A search performed in 2023 may miss structures that were deposited in 2024. If your analysis depends on identifying all structural neighbors of a query, plan to rerun your searches periodically and document the database version for each run.

### Confusing Structural Similarity with Evolutionary Relatedness

Structural similarity does not always imply evolutionary relatedness. Convergent evolution can produce similar structures in proteins that share no common ancestor. Additionally, structural similarity can arise from shared ligand binding or shared environmental constraints instead of shared ancestry. When interpreting DALI results, consider whether the structural similarity is likely to reflect homology or convergence.

The distinction between homology and structural similarity is particularly important in viral classification, where structural conservation can reveal deep evolutionary relationships but can also reflect convergent adaptation to similar functional constraints [8]. Additional evidence, such as sequence similarity in conserved regions or shared functional features, can help distinguish these possibilities.

## Limitations of DALI and Structural Comparison

Understanding the limitations of DALI helps you interpret results appropriately and avoid overinterpreting structural similarities.

### Sensitivity to Conformational Changes

DALI compares static structures, and proteins that undergo significant conformational changes may show lower structural similarity than their evolutionary relationship would suggest. A protein in an open conformation may not align well with its homolog in a closed conformation, even though the underlying fold is the same.

For proteins with multiple conformational states, consider comparing representative structures from each state. The Scipion-EM-ProDy plugin enables analysis of conformational dynamics and can help you understand how structural variation affects comparisons [11]. If your protein of interest is known to undergo conformational changes, interpret DALI results with this flexibility in mind.

### Incomplete Coverage of Structural Space

The PDB contains structures for only a fraction of all proteins, and some structural families are overrepresented while others are underrepresented. If your query protein belongs to a family with few representatives in the PDB, DALI may fail to identify genuine structural neighbors simply because they have not been determined experimentally.

This limitation is particularly relevant for proteins from understudied organisms or proteins with unusual folds. If DALI returns no significant matches, consider whether the absence of matches reflects genuine novelty or simply the lack of related structures in the database.

### Quality Dependence of Input Structures

The quality of your query structure affects the quality of your DALI results. Low-resolution structures, structures with modeling errors, and structures with missing regions can all produce misleading comparisons. For predicted structures, the accuracy of the model determines the reliability of the comparison.

The NCBI provides access to structure validation resources that can help you assess the quality of your query structure before submission [1]. If your structure has regions with poor quality, consider whether those regions should be excluded from the comparison or whether the results should be interpreted with additional caution.

### Computational Cost for Large-Scale Analyses

Running DALI searches for many query structures can be computationally expensive, particularly if each search compares against the entire PDB. For large-scale analyses, consider whether alternative tools with faster search speeds, such as Foldseek, might be more appropriate. The tradeoff between speed and sensitivity should be evaluated in the context of your specific research question.

## Safety and Regulatory Context

Structural comparison using DALI is a computational analysis that does not involve hazardous materials or regulated procedures. However, researchers should be aware of the broader context in which structural information is used, particularly when it informs experimental work or clinical applications.

### Data Management and Privacy

When working with structures that are not yet publicly released, be mindful of data sharing policies. Structures determined by structural genomics consortia or industrial partners may be subject to embargo periods before public release. Ensure that your use of DALI complies with any data use agreements that apply to your structures.

### Responsible Use of Structural Information

Structural comparisons can reveal functional relationships that have implications for drug design, protein engineering, and other applications. When using DALI results to guide experimental work, ensure that your conclusions are supported by sufficient evidence and that you have considered alternative interpretations.

The European Bioinformatics Institute provides training on responsible data use and the ethical considerations of bioinformatics research [2]. Familiarizing yourself with these principles helps ensure that your structural analysis work meets professional standards.

### Reproducibility Standards

Funding agencies and journals increasingly require that computational analyses be reproducible. Maintaining detailed records of your DALI analyses, including parameters and database versions, supports reproducibility and enables others to verify your results. The nf-core documentation provides guidance on reproducibility standards for bioinformatics pipelines that can be adapted to structural comparison work [5].

## Professional Escalation Criteria

Knowing when to seek additional expertise can prevent errors and improve the quality of your structural analysis. The following situations warrant consultation with a structural bioinformatics specialist or a more experienced colleague.

### When Results Are Unexpected

If DALI returns no significant matches for a protein that you expect to have structural homologs, or if it returns matches that contradict established knowledge about your protein, seek a second opinion. Unexpected results may indicate problems with your query structure, errors in your interpretation, or genuine novelty that requires additional investigation.

### When Structural Similarity Is Ambiguous

If the Z-scores for your top matches fall in a gray zone where significance is unclear, consult with someone who has experience interpreting DALI results. Ambiguous results may require additional analyses, such as pairwise comparisons with TM-align or examination of the structural alignments in detail.

### When Results Will Support High-Stakes Conclusions

If your DALI results will be used to support conclusions about protein function, evolutionary relationships, or drug design, have your analysis reviewed by a colleague with structural biology expertise. Independent verification of your methods and interpretation can catch errors that you might have missed.

### When You Need to Integrate Multiple Lines of Evidence

Structural comparison is often one piece of a larger analysis that includes sequence analysis, functional assays, and other data. If you need to integrate DALI results with other types of evidence, consider consulting with a bioinformatics specialist who can help you design an integrated analysis. Tools like MPACT demonstrate how multiple metrics can be combined to provide a more complete picture of protein relationships [8].

## Frequently Asked Questions

### What is the minimum Z-score that indicates significant structural similarity?

A Z-score of 2.0 is commonly used as a threshold for significance in DALI results, but this value is not absolute. The significance of a given Z-score depends on the size of the query structure and the composition of the database being searched. For small proteins or proteins with common folds, higher Z-scores may be needed to achieve the same level of confidence. Always examine the alignment coverage and RMSD alongside the Z-score to assess the quality of a match.

### Can I use DALI to compare predicted protein structures?

Yes, DALI can accept predicted structures, but the results should be interpreted with additional caution. Prediction errors can create artificial structural similarities or obscure genuine ones. If you are comparing predicted structures, validate the quality of your models before submission and consider whether low-confidence regions should be excluded from the comparison. Comparing predicted structures against experimentally determined structures can help assess the accuracy of the predictions.

### How long does a DALI search take?

The time required for a DALI search depends on server load and the complexity of your query structure. Simple queries may complete in minutes, while complex queries or searches performed during periods of high demand may take longer. The server provides a job identifier that you can use to retrieve results when processing is complete. For large-scale analyses, consider whether alternative tools with faster search speeds might be more appropriate.

### What is the difference between DALI Z-scores and TM-scores?

DALI Z-scores measure the statistical significance of structural similarity relative to a background distribution of unrelated proteins. TM-scores measure topological similarity on a normalized scale where scores above 0.5 generally indicate the same fold and scores above 0.7 indicate highly similar structures. Both metrics are useful, but they answer different questions. Z-scores tell you whether a match is statistically significant, while TM-scores tell you how similar the topologies are.

### Why does DALI sometimes fail to detect structural similarity between homologous proteins?

DALI can fail to detect structural similarity for several reasons. Large conformational changes between the compared structures can reduce the alignment score. Proteins that share only a small structural domain may fall below the significance threshold even though the shared domain is biologically meaningful. Additionally, if the homologous protein has not been deposited in the PDB, DALI cannot detect it. Consider these possibilities when interpreting negative results.

### How should I choose between DALI and other structural comparison tools?

The choice of tool depends on your research question. DALI is well suited for discovering structural neighbors in the PDB because its Z-score provides a statistical measure of significance. TM-align is well suited for pairwise comparisons where you want a normalized measure of topological similarity. Foldseek is well suited for very large-scale searches where speed is a priority. For classification tasks that require multiple lines of evidence, integrated tools that combine sequence and structure metrics may be more appropriate [8].

### Can DALI results be used to infer evolutionary relationships?

DALI results can provide evidence for evolutionary relationships, but structural similarity alone does not prove homology. Convergent evolution can produce similar structures in proteins that share no common ancestor. When using DALI results to support evolutionary conclusions, consider additional evidence such as sequence similarity in conserved regions, shared functional features, and phylogenetic context. Structural similarity is most convincing as evidence of homology when combined with other lines of evidence.

### How often should I rerun DALI searches to stay current with the PDB?

The frequency of rerunning DALI searches depends on your research needs. If you are using DALI results to support ongoing research, consider rerunning your searches every six to twelve months to capture newly deposited structures. If your analysis is complete and you have documented the database version used, you may not need to rerun the search unless new structures are likely to change your conclusions. Document the database version for each search to enable comparisons across time.

## Related Bioinformatics Guides

- [Structural Comparison and Alignment Algorithms for Protein 3D Structures](/knowledge/bioinformatics/structural-comparison-and-alignment-algorithms-for-protein-3d-structures)
- [How To Use Alphafold To Predict Structure: Structural Analysis and Computational Methodologies in Bioinformatics](/knowledge/bioinformatics/how-to-use-alphafold-to-predict-structure)
- [Single-Cell Isolation Techniques: A Practical Comparison](/knowledge/bioinformatics/single-cell-isolation-techniques-a-practical-comparison)
- [How to Interpret Gene Set Enrichment Analysis Results](/knowledge/bioinformatics/how-to-interpret-gene-set-enrichment-analysis-results)
- [Protein Language Models in Bioinformatics: A Practical Guide to Selection and Application](/knowledge/bioinformatics/protein-language-models-in-bioinformatics-a-practical-guide-to-selection-and-application)

## References and Further Reading

- [NCBI Data Resources](https://www.ncbi.nlm.nih.gov/). National Center for Biotechnology Information.
- [EMBL-EBI Training](https://www.ebi.ac.uk/training). European Bioinformatics Institute.
- [Bioconductor](https://bioconductor.org/). Bioconductor Project.
- [Galaxy Training Network](https://training.galaxyproject.org/). Galaxy Project.
- [nf-core Documentation](https://nf-co.re/docs). nf-core.
- [The Carpentries Lessons](https://carpentries.org/lessons). The Carpentries.
- [Alishewanella Phage LSH1 from the Sea Surface Microlayer Provides a Novel Minimalistic View of the Siphoviral Hub Structure](https://europepmc.org/article/PMC/PMC13243795)
- [Integrating Sequence- and Structure-Based Similarity Metrics for the Demarcation of Multiple Viral Taxonomic Levels.](https://doi.org/10.3390/v17050642). 2025.
- [Prenatal nicotine exposure induces epigenetic alterations in the Notch signaling genes in the proximal colon in rats.](https://doi.org/10.1186/s12864-025-11960-9). 2025.
- [Scanning sample-specific miRNA regulation from bulk and single-cell RNA-sequencing data.](https://doi.org/10.1186/s12915-024-02020-x). 2024.
- [Scipion-EM-ProDy: A Graphical Interface for the ProDy Python Package within the Scipion Workflow Engine Enabling Integration of Databases, Simulations and Cryo-Electron Microscopy Image Processing.](https://doi.org/10.3390/ijms241814245). 2023.

> This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.