# How to Use SIFTS to Map UniProt Annotations onto PDB Structures: A Practical Tutorial for Cross-Database Integration Researchers who need to connect sequence-level annotations from UniProt with three-dimensional coordinates in the Protein Data Bank face a recurring problem: the two databases use different residue numbering systems, different chain identifiers, and different granularity for describing functional sites. The SIFTS resource, maintained by the Protein Data Bank in Europe, provides a residue-level mapping between UniProt and PDB entries that resolves these discrepancies. This tutorial explains how to access SIFTS data, interpret the mapping files, and apply them to common research tasks such as locating disease-associated variants on structures, interpreting docking results, and validating structure-based functional predictions. ## The Problem of Cross-Database Residue Mapping UniProt entries describe proteins using the full amino acid sequence of the canonical isoform, with numbering that starts at the initiator methionine. PDB entries describe the experimentally determined structure, which often contains only a fragment of the full protein, may include expression tags or affinity purification sequences, and uses numbering that reflects the construct used for crystallization or cryo-EM analysis. A residue that appears at position 245 in UniProt might appear at position 198 in the PDB file, or might be absent entirely because that region was disordered or cleaved during sample preparation. The practical consequence is that a researcher who identifies a pathogenic variant at UniProt position 245 cannot simply open the PDB file and look at residue 245. The structural context of that variant, including its proximity to active site residues, binding interfaces, or post-translational modification sites, requires an accurate residue-level correspondence between the two numbering systems. SIFTS provides this correspondence through a curated, weekly-updated mapping that aligns each PDB residue to its equivalent UniProt residue where such an equivalence exists. The need for such mapping extends beyond variant interpretation. Molecular docking studies often produce predicted binding poses that must be interpreted in the context of known functional residues. Structure-based drug design requires knowledge of which structural regions correspond to annotated domains or binding sites. Comparative analysis of multiple structures of the same protein requires a common reference frame for residue numbering. Each of these tasks depends on reliable sequence-structure mapping. Protein structures are central to understanding the function, mechanism, and disease-causing variants of proteins within any living cell. Experimental techniques such as X-ray crystallography, cryo-electron microscopy, and NMR spectroscopy are employed by researchers to determine such structures. Through structure inspection in molecular viewers, combined with supporting biochemical and biophysical experiments, scientists are able to identify the function of a protein, its reaction mechanism, and effects caused by sequence variation. These detailed findings, supported by experimental results, are documented by being described in the scientific literature and by making the accompanying data open source. However, it has become increasingly difficult for a reader, in particular a non-expert, to access the correct additional information and assess the validity of the conclusions drawn based on experimental results. A reader is often required to resort to a number of different software packages to access the different data types. Recent developments include artificial intelligence and text-mining-supported software tools that allow the association of mentions in the text of one or more specific protein residues with their corresponding counterparts in the respective protein structure or structures. Such applications allow a researcher to explore a residue of interest in the context of a publication and its respective protein structure, supported by its experimental evidence, in a single view [7]. ## What SIFTS Provides SIFTS, which stands for Structure Integration with Function, Taxonomy and Sequences, is a collaborative effort that produces residue-level mappings between PDB structures and UniProt sequences. The resource also integrates taxonomic information and functional annotations from other databases, but the core utility for most researchers is the residue-level sequence mapping. The mapping is generated through a combination of sequence alignment and manual curation. For each PDB entry, the sequence of each chain is aligned to the corresponding UniProt sequence. The alignment accounts for insertions, deletions, and sequence differences that may arise from expression constructs, natural variants, or crystallization artifacts. The output is a residue-by-residue correspondence that specifies, for each PDB residue, the equivalent UniProt residue number and the confidence of that assignment. SIFTS data are updated weekly to reflect new PDB entries and revisions to UniProt sequences. The mapping files are available in several formats, including XML and flat files, and can be downloaded in bulk or accessed through the PDBe API. The European Bioinformatics Institute provides training materials that cover the use of PDBe and related resources, which can help researchers understand the data formats and access methods [2]. The integration of sequence and structural data is a growing area of bioinformatics. Resources such as PhyreRisk demonstrate the value of bridging genomic, proteomic, and structural data to facilitate the mapping of human variants onto protein structures. PhyreRisk provides information on 20,214 human canonical proteins and an additional 22,271 alternative protein sequences (isoforms), with structural coverage for 70 percent of the human proteome by storing experimental structures and pre-built models. Such resources empower researchers to translate genetic data into protein structural information, thereby providing a more comprehensive appreciation of the functional impact of variants [8]. ## At a Glance: SIFTS Mapping Workflow | Step | Action | Output | Common Error | |------|--------|--------|--------------| | 1 | Identify the UniProt accession for your protein of interest | Stable protein identifier | Using a gene name or obsolete accession | | 2 | Retrieve the SIFTS mapping file for the PDB entry or UniProt accession | XML or flat file with residue correspondences | Downloading the wrong chain or isoform | | 3 | Parse the mapping to extract residue equivalences | Table of PDB residue to UniProt residue pairs | Ignoring mapping confidence scores | | 4 | Apply the mapping to your annotation set | Structural coordinates for each annotated residue | Assuming all residues have structural coverage | | 5 | Validate the mapping against known landmarks | Confirmed correspondence for active site or conserved residues | Accepting mappings without checking sequence identity | ## Core Principles of SIFTS Mapping ### Residue-Level Correspondence The fundamental unit of SIFTS mapping is the individual amino acid residue. Each PDB residue is assigned to a specific UniProt residue when a confident alignment exists. This residue-level granularity distinguishes SIFTS from coarser mapping approaches that might only associate a PDB chain with a UniProt entry without specifying which residues correspond to which. The residue-level mapping is essential for tasks such as variant mapping. When a researcher has a list of disease-associated variants with UniProt positions, the SIFTS mapping allows each variant position to be located in the three-dimensional structure. The structural context, including secondary structure, solvent accessibility, and proximity to functional sites, can then be assessed. ### Chain-Level Context PDB entries often contain multiple chains, and these chains may correspond to different UniProt entries or to different regions of the same UniProt entry. SIFTS mapping is performed per chain, and the mapping file specifies which UniProt accession corresponds to each PDB chain. This chain-level context is critical for interpreting mappings in multimeric complexes or in structures that contain multiple domains from the same protein. ### Isoform Awareness UniProt entries can have multiple isoforms, and the canonical sequence may not be the one that was used for structure determination. SIFTS mapping accounts for this by specifying which UniProt isoform corresponds to the PDB chain. Researchers who are working with a specific isoform need to verify that the SIFTS mapping uses the same isoform, or they need to adjust their residue numbering accordingly. ### Mapping Confidence Not all residue correspondences are equally reliable. SIFTS provides confidence information that reflects the quality of the sequence alignment at each position. Regions with high sequence identity and no insertions or deletions typically have high confidence, while regions with sequence divergence or alignment gaps may have lower confidence. Researchers should examine confidence scores when interpreting mappings, particularly for residues in variable regions or at the termini of constructs. ## Accessing SIFTS Data ### Bulk Download The most straightforward approach for researchers who need to map a small number of proteins is to download the SIFTS mapping file for specific PDB entries. The PDBe website provides access to SIFTS data through entry pages and through the REST API. For each PDB entry, the SIFTS XML file contains the residue-level mapping for all chains. For large-scale projects that involve mapping many structures, bulk download of the complete SIFTS dataset is appropriate. The full dataset is available as a compressed archive that can be downloaded and processed locally. This approach requires substantial storage and computational resources but provides the flexibility to query the data without repeated network requests. ### Programmatic Access The PDBe REST API allows programmatic retrieval of SIFTS data. Researchers can query the API for specific PDB entries or UniProt accessions and receive the mapping data in a structured format. This approach is suitable for integration into analysis pipelines and for automated quality checks. The National Center for Biotechnology Information provides a range of sequence analysis resources that can complement SIFTS data [1]. For example, BLAST searches can be used to verify sequence relationships between PDB chains and UniProt entries, and the NCBI databases can provide additional context about genes, transcripts, and variants. ### File Formats SIFTS data are available in XML format, which provides a hierarchical representation of the mapping. The XML structure includes entries for each PDB chain, with residue-level elements that specify the PDB residue number, the UniProt residue number, and confidence information. Flat file formats are also available for researchers who prefer tabular data. The choice of format depends on the intended use. XML is more expressive and can represent complex relationships, but requires parsing. Flat files are easier to load into spreadsheet applications or data analysis tools but may lose some of the hierarchical structure. ## Practical Workflow for Mapping UniProt Annotations ### Step 1: Define Your Annotation Set Before accessing SIFTS data, clearly define the set of UniProt annotations that need to be mapped. These annotations might include: - Disease-associated variants from ClinVar or other databases - Post-translational modification sites - Active site residues from catalytic annotations - Binding site residues for ligands or other proteins - Domain boundaries from InterPro or Pfam - Residues identified in mutagenesis experiments Each annotation should be associated with a specific UniProt accession and residue position. The HuVarBase resource provides an example of how human variant data can be integrated with sequence and structural features, including domain information and post-translational modification sites. HuVarBase assimilates publicly available human variant data at protein level and gene level into a comprehensive resource, integrating amino acid sequence, secondary structure of the mutant residue, domain, function, subcellular location and post-translational modification with gene level data such as gene name, chromosome number and genome position, DNA mutation, mutation type origin and rs ID number [11]. Researchers can use such resources to assemble their annotation sets. ### Step 2: Identify Relevant PDB Entries For each UniProt accession, identify the PDB entries that contain the corresponding protein. The PDBe website provides a mapping from UniProt accessions to PDB entries, and this information is also available through the SIFTS data. Researchers should consider: - Which structures cover the region of interest - The resolution or quality of each structure - Whether the structure contains relevant ligands or binding partners - The biological context of the structure, such as apo versus holo forms or different conformational states For proteins with many structures, a systematic approach to selecting representative structures is needed. The choice of structure can significantly affect the interpretation of mapped annotations, particularly for flexible regions or conformational changes. ### Step 3: Retrieve SIFTS Mapping Files Download the SIFTS mapping files for the selected PDB entries. If using the PDBe REST API, construct the appropriate queries to retrieve the mapping data. For bulk analysis, download the complete SIFTS dataset and filter locally. ### Step 4: Parse and Validate the Mapping Parse the SIFTS files to extract the residue-level correspondences. For each PDB chain, build a mapping table that specifies the PDB residue number, the UniProt residue number, and the confidence score. Validate the mapping by checking: - The UniProt accession matches your protein of interest - The sequence identity between the PDB chain and the UniProt sequence is consistent with the expected relationship - Known landmarks, such as active site residues or conserved motifs, map to the expected positions The Galaxy Training Network provides accessible tutorials on bioinformatics workflows that can help researchers develop reproducible analysis pipelines for tasks such as parsing and validating mapping data [4]. ### Step 5: Apply the Mapping to Your Annotations For each UniProt annotation, use the mapping table to identify the corresponding PDB residue. If the annotation falls in a region that is not covered by the structure, note this as a limitation. If the annotation falls in a region with low mapping confidence, verify the correspondence using alternative methods. ### Step 6: Visualize and Interpret Load the structure and the mapped annotations into a molecular viewer. Color the structure to highlight the annotated residues and examine their structural context. Consider: - Secondary structure context, including helix, strand, or loop - Solvent accessibility - Proximity to ligands or binding interfaces - Interactions with neighboring residues - Conservation across related structures The PhyreRisk application provides an example of how sequence-structure mapping can be used to interpret human genetic variants in a structural context. PhyreRisk allows the user to explore novel variants using genomic coordinates formats including Ensembl, VCF, reference SNP ID and HGVS notations, and supports mapping variants using amino acid coordinates and searching for genes or proteins of interest [8]. While PhyreRisk uses its own mapping approach, the underlying principle of connecting sequence annotations to structural features is the same. ## Options and Tradeoffs in Mapping Approaches ### SIFTS versus Direct Sequence Alignment SIFTS provides a curated mapping that has been validated by the PDBe team. Direct sequence alignment using tools such as BLAST or Clustal Omega can produce similar mappings but requires careful parameter selection and validation. The advantage of SIFTS is that the mapping has been reviewed and corrected where necessary, reducing the risk of errors from automated alignment. The tradeoff is that SIFTS may not cover all PDB entries or all UniProt isoforms. For structures that are not included in SIFTS, or for non-canonical isoforms, direct sequence alignment may be the only option. Researchers should be aware of the coverage limitations and verify that their protein of interest is included in SIFTS. ### SIFTS versus Structure-Based Superposition For proteins with multiple structures, structure-based superposition can be used to transfer annotations between structures. This approach aligns the three-dimensional coordinates of the structures and identifies equivalent residues based on spatial proximity. Structure-based superposition is particularly useful when the sequence identity between structures is low but the fold is conserved. The tradeoff is that structure-based superposition requires structures to be aligned, which can be computationally intensive and may produce ambiguous results in flexible regions. SIFTS provides a sequence-based mapping that is independent of structural alignment and can be applied consistently across all structures. ### SIFTS versus Variant-Specific Tools Several tools provide variant mapping as a specific service. PhyreRisk, for example, allows users to map human variants onto protein structures and provides pre-computed structural coverage for a large fraction of the human proteome. PhyreRisk reports 55,732 experimentally, multi-validated protein interactions from IntAct and 24,260 experimental structures of protein complexes [8]. These tools can be convenient for researchers who work primarily with human variants and do not need the full flexibility of SIFTS. The tradeoff is that variant-specific tools may not support all organisms, all types of annotations, or all PDB entries. SIFTS provides a general-purpose mapping that can be applied to any protein with a UniProt entry and a PDB structure. ## Observations and Measurements for Quality Control ### Sequence Identity Checks A fundamental quality check for SIFTS mapping is to verify the sequence identity between the PDB chain and the UniProt sequence. For most structures, the sequence identity should be high, typically above 90 percent, with differences concentrated in expression tags, linker regions, or engineered mutations. Low sequence identity may indicate that the wrong UniProt accession was used or that the PDB chain corresponds to a different isoform. ### Coverage Assessment Assess the fraction of the UniProt sequence that is covered by the PDB structure. Complete coverage is uncommon, and most structures cover only a portion of the full protein. The coverage assessment should consider: - Which regions of the protein are present in the structure - Which regions are missing due to disorder or construct design - Whether the missing regions are relevant to the annotations being mapped For variant interpretation, the key question is whether the variant position falls within the structurally covered region. Variants in disordered regions or in regions that were not included in the construct cannot be interpreted structurally. ### Confidence Score Distribution Examine the distribution of confidence scores across the mapping. High-confidence mappings should dominate, with low-confidence mappings concentrated in specific regions such as termini or variable loops. A mapping with many low-confidence positions may indicate alignment problems or sequence divergence that requires investigation. ### Landmark Validation Identify residues that are known to be functionally important and verify that they map to the expected positions. For example, if the active site of an enzyme is known to contain a specific catalytic residue, confirm that this residue maps to a position in the structure that is consistent with the active site geometry. Landmark validation provides an independent check on the mapping quality. ## Records and Documentation for Reproducibility ### Recording Mapping Parameters For reproducible analysis, record the specific parameters used for mapping: - SIFTS data version or download date - UniProt accession and isoform - PDB entry and chain identifier - Software and version used for parsing - Any filtering criteria applied to the mapping The nf-core documentation emphasizes the importance of reproducibility in bioinformatics workflows and provides standards for pipeline documentation and configuration [5]. Applying similar standards to SIFTS mapping ensures that results can be reproduced and verified. ### Documenting Coverage Limitations For each mapping analysis, document the structural coverage of the protein and the regions that could not be mapped. This documentation is essential for interpreting negative results. If a variant cannot be mapped to a structure, the reason should be recorded: the region is disordered, the construct does not include the region, or the mapping confidence is too low. ### Version Control for Analysis Scripts Maintain version control for any scripts used to parse SIFTS data and apply mappings. The Carpentries provides lessons on version control with Git that are directly applicable to bioinformatics analysis workflows [6]. Version control ensures that analysis scripts can be traced to specific versions and that changes can be reviewed and documented. ## Common Failure Patterns in SIFTS Mapping ### Isoform Mismatch A frequent error is using the wrong UniProt isoform. The PDB structure may correspond to an isoform that differs from the canonical sequence, and the residue numbering will be shifted accordingly. Researchers should verify that the SIFTS mapping specifies the correct isoform and adjust their annotation coordinates if necessary. ### Chain Confusion PDB entries with multiple chains can cause confusion when the chains correspond to different proteins or to different regions of the same protein. The SIFTS mapping file specifies the UniProt accession for each chain, and researchers must ensure that they are using the correct chain for their analysis. ### Expression Tag Contamination Many recombinant proteins are expressed with tags that are not part of the native sequence. These tags appear in the PDB sequence but are not present in the UniProt entry. SIFTS mapping accounts for these tags, but researchers who are manually inspecting structures may be confused by the extra residues at the termini. ### Construct Boundary Artifacts The boundaries of the construct used for structure determination may not align with domain boundaries or functional regions. Residues near the construct boundaries may have different conformations or interactions than they would in the full-length protein. Researchers should be cautious when interpreting mapped annotations near construct boundaries. ### Low-Complexity Regions Low-complexity regions, such as polyproline tracts or glycine-rich loops, can be difficult to align unambiguously. SIFTS mapping may have lower confidence in these regions, and the residue-level correspondence may be uncertain. Researchers should treat mappings in low-complexity regions with caution. ## Limitations of SIFTS Mapping ### Coverage Gaps SIFTS does not cover all PDB entries or all UniProt accessions. Some structures may not have been processed by SIFTS, and some UniProt entries may not have corresponding structures. Researchers should check coverage before relying on SIFTS for their analysis. ### Isoform Limitations SIFTS mapping is primarily based on canonical UniProt sequences. Non-canonical isoforms may not have direct SIFTS mappings, and researchers working with specific isoforms may need to use alternative approaches or manually adjust the mapping. ### Dynamic Regions Regions of proteins that are disordered or that adopt multiple conformations may not be well represented in crystal structures. SIFTS mapping can identify the sequence positions of these regions, but the structural interpretation is limited by the absence of ordered electron density. ### Engineered Mutations Structures that contain engineered mutations, such as stabilizing mutations or active site mutations, will have sequence differences from the wild-type UniProt sequence. SIFTS mapping accounts for these differences, but researchers should be aware that the structural context may differ from the wild-type protein. ## Safety and Regulatory Context for Structural Interpretation ### Clinical Variant Interpretation When SIFTS mapping is used to interpret clinically relevant variants, the structural information should be considered as one line of evidence among many. Structural context can suggest mechanisms of pathogenicity, such as disruption of a binding interface or destabilization of the protein core, but structural predictions should not be used as the sole basis for clinical decisions. The HuVarBase resource integrates variant data with sequence and structural features to support disease-related analysis. HuVarBase contains 774,863 variant records that can be searched with options to display, visualize and download the results [11]. Researchers who are interpreting human variants should consult multiple resources and consider the full range of evidence, including population frequency, functional assays, and clinical observations. ### Reproducibility Standards For research applications, the reproducibility of SIFTS mapping is essential. The weekly updates to SIFTS mean that mappings can change over time as new structures are added or as UniProt sequences are revised. Researchers should record the date and version of the SIFTS data used in their analysis and re-run analyses if the underlying data have changed. The Bioconductor project provides tools and workflows for reproducible genomic analysis that can be adapted for structural bioinformatics [3]. Using versioned packages and documented workflows ensures that analyses can be reproduced by other researchers. ### Professional Escalation Criteria When SIFTS mapping produces unexpected results, or when the mapping is needed for high-stakes decisions, professional escalation may be appropriate. Consider escalating to a structural biology expert or to the PDBe help desk when: - The mapping confidence is consistently low across a protein - The sequence identity between the PDB chain and UniProt sequence is unexpectedly low - The mapping produces results that contradict established knowledge about the protein - The interpretation of a clinically relevant variant depends critically on the mapping ## Practical Implementation Steps ### Setting Up a SIFTS Mapping Pipeline 1. Install the necessary software for downloading and parsing SIFTS data. Python with the requests library and an XML parser is a common choice. 2. Write a script that takes a UniProt accession and a list of residue positions as input and returns the corresponding PDB residues for a specified structure. 3. Test the script on a protein with known structure and known functional residues to verify that the mapping is correct. 4. Document the script and its parameters for reproducibility. ### Integrating SIFTS with Structure Analysis Tools SIFTS mapping can be integrated with molecular visualization tools to display UniProt annotations on structures. The workflow involves: 1. Loading the PDB structure into the visualization tool 2. Loading the SIFTS mapping data 3. Selecting residues based on UniProt positions 4. Coloring or highlighting the selected residues 5. Generating images or animations for publication ### Automating Quality Checks For large-scale mapping projects, automate the quality checks described above. The automated checks should verify: - UniProt accession consistency - Sequence identity thresholds - Coverage of the region of interest - Confidence score distributions - Landmark residue validation The Galaxy Training Network provides examples of how to build automated bioinformatics workflows that include quality control steps [4]. These principles can be applied to SIFTS mapping pipelines. ## Interpreting Mapped Annotations in Structural Context ### Variant Effect Prediction When a disease-associated variant is mapped to a structure, the structural context can suggest potential mechanisms of pathogenicity. A variant that maps to the protein core may destabilize the fold, while a variant that maps to a binding interface may disrupt protein-protein interactions. The structural interpretation should be combined with other evidence, including evolutionary conservation, biochemical data, and clinical observations. The PhyreRisk application demonstrates how structural mapping can be used to interpret human genetic variants, providing structural coverage for a large fraction of the human proteome. PhyreRisk is designed to empower researchers to translate genetic data into protein structural information, thereby providing a more comprehensive appreciation of the functional impact of variants [8]. Researchers can use such tools to generate hypotheses about variant mechanisms that can be tested experimentally. ### Docking Interpretation Molecular docking studies produce predicted binding poses that need to be interpreted in the context of known functional residues. SIFTS mapping allows the docking results to be compared with annotated binding sites, catalytic residues, or post-translational modification sites. This comparison can validate docking predictions or identify inconsistencies that require further investigation. ### Domain and Functional Site Analysis SIFTS mapping can be used to locate domain boundaries and functional sites on structures. By mapping InterPro or Pfam domain annotations to the structure, researchers can visualize the spatial organization of domains and their relationships to functional features. This analysis is particularly useful for multi-domain proteins where the domain organization is not obvious from the sequence alone. ### Cross-Structure Comparison For proteins with multiple structures, SIFTS mapping provides a common reference frame for comparison. By mapping all structures to the same UniProt sequence, researchers can compare the positions of functional residues across different conformational states, different ligand-bound forms, or different species orthologs. ## Integrating SIFTS with Other Bioinformatics Resources ### Combining SIFTS with Variant Databases Human variant databases could be better exploited if the variant data available in multiple resources is integrated in a single comprehensive resource along with sequence and structural features. Such integration would improve the analyses of variants for disease prediction, prevention or treatment. The HuVarBase assimilates publicly available human variant data at protein level and gene level into a comprehensive resource, integrating protein level data such as amino acid sequence, secondary structure of the mutant residue, domain, function, subcellular location and post-translational modification with gene level data such as gene name, chromosome number and genome position, DNA mutation, mutation type origin and rs ID number [11]. When combined with SIFTS mapping, such variant databases can provide a direct path from genomic variation to structural context. ### Using SIFTS in Metagenomic and Microbiome Studies Microbial communities and their associated bioactive compounds are often disrupted in conditions such as the inflammatory bowel diseases. In well-characterized environments such as the human gastrointestinal tract, more than one-third of microbial proteins are uncharacterized and often expected to be bioactive. Systematic identification of protein families as potentially bioactive with respect to gut inflammation during IBD has revealed that about half of these families have not been functionally characterized previously on the basis of homology or experiment. Predictions from high-throughput data were validated using targeted experiments that revealed the differential immunogenicity of prioritized Enterobacteriaceae pilins and the contribution of homologues of von Willebrand factors to the formation of Bacteroides biofilms in a manner dependent on mucin levels [9]. SIFTS mapping can play a role in such studies by connecting predicted functional residues in microbial proteins to experimental structures where available. ### Spatial Transcriptomics and Structural Context Dysregulation of communication between cells mediates complex diseases such as cancer and diabetes, and detecting cell-cell communication at scale remains one of the greatest challenges in transcriptomics. Computational approaches such as CellNEST have been developed to decipher patterns of communication, introducing relay-network communication detection that identifies putative ligand-receptor-ligand-receptor communication. CellNEST detects T cell homing signals in human lymph nodes, identifies aggressive cancer communication in lung adenocarcinoma and colorectal cancer, and predicts new patterns of communication that may act as relay networks in pancreatic cancer [10]. When such transcriptomics analyses identify specific ligand-receptor pairs, SIFTS mapping can be used to locate the relevant binding interfaces on experimental structures, providing structural context for the predicted communication events. ## Troubleshooting Common SIFTS Mapping Problems ### Missing Residues in the Mapping When residues are missing from the SIFTS mapping, the cause is often one of the following: - The residue falls in a disordered region that was not modeled in the structure - The residue is in a region that was not included in the expression construct - The residue is in a region of low sequence identity where the alignment is ambiguous - The residue is part of an insertion that has no equivalent in the UniProt sequence For each missing residue, document the reason and consider whether alternative structures might provide coverage of the region. ### Unexpected Residue Numbering If the residue numbering in the PDB file does not match the expected UniProt numbering, check whether the structure contains expression tags or affinity purification sequences. These additional residues shift the numbering of all subsequent residues. SIFTS mapping accounts for these shifts, but manual inspection of the PDB file may be confusing. ### Multiple UniProt Accessions for One Chain In some cases, a single PDB chain may map to multiple UniProt accessions. This situation can arise when the construct contains a fusion protein or when the sequence matches multiple closely related isoforms. The SIFTS mapping file will specify the primary UniProt accession, but researchers should verify that this accession matches their protein of interest. ### Low Confidence in Specific Regions Low confidence scores in specific regions may indicate alignment ambiguity or sequence divergence. For residues in these regions, consider using direct sequence alignment to verify the correspondence. If the direct alignment agrees with the SIFTS mapping, the confidence can be upgraded. If the direct alignment disagrees, investigate the cause of the discrepancy. ## Best Practices for SIFTS Data Management ### Maintaining a Local Cache For projects that require repeated access to SIFTS data, maintain a local cache of the mapping files. This approach reduces network requests and ensures that the analysis uses a consistent version of the data. The cache should be updated on a regular schedule to incorporate new PDB entries and UniProt revisions. ### Tracking Data Versions Record the version and date of the SIFTS data used for each analysis. This information is essential for reproducing results and for understanding discrepancies between analyses performed at different times. The weekly update cycle means that mappings can change, and analyses that depend on specific mappings should be re-validated periodically. ### Documenting Analysis Decisions For each mapping analysis, document the decisions made during the process: - Which PDB entries were selected and why - Which chains were used for the analysis - How missing residues were handled - How low-confidence mappings were treated - Which validation checks were performed This documentation supports reproducibility and provides context for interpreting the results. ## Advanced Applications of SIFTS Mapping ### Mapping Post-Translational Modifications Post-translational modification sites annotated in UniProt can be mapped to structures using SIFTS. The structural context of a modification site, including solvent accessibility and proximity to binding interfaces, can provide insight into the functional role of the modification. For example, a phosphorylation site that maps to a protein-protein interaction interface may regulate binding through steric or electrostatic effects. ### Analyzing Disease Variant Clusters When multiple disease-associated variants map to the same structural region, this clustering can suggest a common mechanism of pathogenicity. SIFTS mapping enables the identification of such clusters by placing variants in their structural context. The analysis can reveal whether variants cluster at binding interfaces, in the protein core, or at allosteric sites. ### Comparing Orthologous Structures For proteins from different species, SIFTS mapping provides a common reference frame for comparing structures. By mapping each structure to the corresponding UniProt sequence, researchers can identify conserved structural features and species-specific differences. This analysis is particularly useful for understanding the structural basis of species-specific functions or drug sensitivities. ### Validating Predicted Structures SIFTS mapping can be used to validate predicted structures by comparing the mapped positions of known functional residues with the predicted coordinates. If a predicted structure places a catalytic residue far from the expected active site geometry, this discrepancy may indicate a problem with the prediction. Conversely, agreement between predicted and experimental structures at functional residues increases confidence in the prediction. ## Frequently Asked Questions ### How often is SIFTS data updated? SIFTS data are updated weekly to reflect new PDB entries and revisions to UniProt sequences. Researchers should record the date of the SIFTS data used in their analysis and re-run analyses if the underlying data have changed. The weekly update cycle means that mappings can change over time, and analyses that depend on specific mappings should be re-validated periodically. ### What is the difference between SIFTS and PDBe? SIFTS is a specific data resource that provides residue-level mappings between PDB and UniProt. PDBe is the Protein Data Bank in Europe, which maintains SIFTS and provides access to the data through its website and API. PDBe also provides many other resources and tools for structural biology, including training materials and documentation [2]. ### Can SIFTS map annotations to AlphaFold models? SIFTS is primarily designed for experimental structures in the PDB. AlphaFold models are not part of the PDB and are not included in SIFTS mapping. However, the same principles of sequence-structure mapping can be applied to AlphaFold models using direct sequence alignment or by using the residue numbering that is inherent in the model. ### How do I handle proteins with multiple isoforms? For proteins with multiple isoforms, verify that the SIFTS mapping uses the isoform that matches your protein of interest. The SIFTS mapping file specifies the UniProt accession and isoform for each PDB chain. If your isoform of interest is not the one used in the structure, you may need to adjust the residue numbering or use an alternative mapping approach. ### What should I do if a residue is not covered by the structure? If a residue of interest is not covered by the structure, first check whether the residue falls in a region that is disordered or was not included in the construct. If the residue is in a region that is present in the structure but not mapped by SIFTS, investigate the reason for the missing mapping. In some cases, the residue may be in a region of low sequence identity or alignment ambiguity. ### How can I validate a SIFTS mapping? Validate SIFTS mappings by checking sequence identity between the PDB chain and the UniProt sequence, examining the confidence score distribution, and verifying that known functional residues map to expected positions. Independent validation can be performed using direct sequence alignment or by comparing multiple structures of the same protein. ### What are the limitations of using SIFTS for variant interpretation? SIFTS mapping provides the structural context for variants but does not predict the functional consequences of variants. The structural interpretation should be combined with other evidence, including evolutionary conservation, population frequency, functional assays, and clinical observations. Structural predictions should not be used as the sole basis for clinical decisions. ### Where can I find training on structural bioinformatics tools? The European Bioinformatics Institute provides training materials for bioinformatics data resources, including structural biology tools [2]. The Galaxy Training Network offers accessible tutorials on bioinformatics workflows [4], and The Carpentries provides foundational lessons on computing and data analysis [6]. These resources can help researchers develop the skills needed for SIFTS mapping and structural analysis. ## Related Bioinformatics Guides - [Multi-Omics Integration: A Practical Guide to Combining Data Types](/knowledge/bioinformatics/multi-omics-integration-a-practical-guide-to-combining-data-types) - [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) - [Gene Set Enrichment Analysis in R: A Practical Tutorial for Interpreting Omics Data](/knowledge/bioinformatics/gene-set-enrichment-analysis-in-r-a-practical-tutorial-for-interpreting-omics-data) - [Metagenomic Assembly and Binning: A Practical Workflow for Recovering Genomes from Complex Microbial Communities](/knowledge/bioinformatics/metagenomic-assembly-and-binning-a-practical-workflow-for-recovering-genomes-from-complex-microb) - [Multi-Omics Integration: A Practical Workflow for Combining Proteomics, Metabolomics, and Epigenomics Data](/knowledge/bioinformatics/multi-omics-integration-a-practical-workflow-for-combining-proteomics-metabolomics-and-epigenomi) ## 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. - [Associating protein residues in the literature with structural data.](https://doi.org/10.1107/s2059798326000021). 2026. - [PhyreRisk: A Dynamic Web Application to Bridge Genomics, Proteomics and 3D Structural Data to Guide Interpretation of Human Genetic Variants.](https://doi.org/10.1016/j.jmb.2019.04.043). 2019. - [Discovery of bioactive microbial gene products in inflammatory bowel disease.](https://doi.org/10.1038/s41586-022-04648-7). 2022. - [CellNEST reveals cell-cell relay networks using attention mechanisms on spatial transcriptomics.](https://doi.org/10.1038/s41592-025-02721-3). 2025. - [HuVarBase: A human variant database with comprehensive information at gene and protein levels.](https://doi.org/10.1371/journal.pone.0210475). 2019. > This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.