# Incorporating Genomic Variants into Proteomics Searches: How to Detect and Validate Variant Peptides Standard proteomics database searches compare tandem mass spectrometry spectra against reference protein sequences. These reference databases typically contain only canonical protein isoforms and do not include amino acid changes arising from single nucleotide variants (SNVs) or small insertions and deletions (indels). When a sample carries a genomic variant that alters protein sequence, peptides containing that variant will not match any entry in the reference database and will remain unidentified. This article explains how to construct variant-aware protein databases from genomic data, how to integrate them into proteomics search workflows, and how to apply rigorous validation criteria so that variant peptide identifications are trustworthy. The intended reader is a researcher, laboratory professional, or graduate student who already performs standard proteomics database searches and wants to extend their workflow to detect protein-level consequences of genomic variation. The practical outcome is a reproducible pipeline that generates a variant protein database, searches it alongside the reference database, applies false discovery rate control, and validates candidate variant peptides using orthogonal evidence. ## The Problem with Reference-Only Protein Databases Reference protein databases such as those derived from genome annotations contain one representative protein sequence per gene locus. These sequences are useful for identifying the majority of peptides in a typical proteomics experiment, but they are incomplete in a specific and consequential way. They do not systematically include protein variants that arise from genomic polymorphisms or somatic mutations. The consequence is straightforward. If a peptide spectrum is generated from a protein that carries a single amino acid substitution, the spectrum will not match the reference peptide sequence. The search engine will either fail to assign the spectrum or will assign it to a different, incorrect peptide. Variant peptides are therefore systematically missing from standard search results. Evidence from a multi-tissue proteogenomics study illustrates the scale of this challenge. In an analysis of 29 healthy human tissues, researchers identified 9,848 amino acid variants by exome sequencing but could confidently detect only 238 of those variants at the protein level. The authors noted that proteogenomics remains challenging, needs better computational methods, and requires rigorous validation. This finding underscores two points. First, variant peptides are detectable in real samples. Second, detection is difficult enough that careful workflow design and validation are essential. The same principle applies across species. A researcher studying livestock, crops, or model organisms faces the same limitation. If the reference proteome for the species does not contain the variant allele present in the sample, variant peptides will be missed. ## What Counts as a Variant Peptide A variant peptide is a peptide whose amino acid sequence differs from the reference sequence due to a genomic variant. The most common class is a single amino acid variant (SAAV) caused by a nonsynonymous SNV. A single nucleotide change in the coding sequence leads to a single amino acid substitution in the protein. The resulting peptide contains that substituted residue and will not match the reference peptide. Indels are a second class. A small insertion or deletion in the coding sequence that does not shift the reading frame will add or remove amino acids in the protein. An indel that shifts the reading frame produces a completely altered sequence downstream of the variant site, often including a premature stop codon. Peptides from frameshifted regions are entirely absent from the reference database. A third class involves splice variants. Alternative splicing generates mRNA isoforms that differ in exon composition. Peptides spanning novel exon-exon junctions are not present in the reference proteome if the reference database contains only the canonical transcript. A large-scale analysis of alternative splicing across 32 cancer types from 8,705 patients identified approximately 930 exon-exon junctions per tumor that were not typically found in normal samples. The same study confirmed approximately 1.7 neojunction-derived peptides and 0.6 SNV-derived peptides per tumor sample in Clinical Proteomic Tumor Analysis Consortium data for breast and ovarian tumors. These numbers are modest per sample, which reinforces the need for sensitive and specific detection methods. For the purposes of this article, the workflow described below focuses primarily on SNVs and small indels, with a section on incorporating splice variants where relevant. ## At a Glance: Variant Peptide Detection Workflow The table below summarizes the key decisions and outputs at each stage of a variant peptide detection workflow. Use this table as a planning reference before you begin database construction. | Workflow Stage | Primary Input | Key Decision | Typical Output | | --- | --- | --- | --- | | Variant calling | Raw sequencing reads or aligned BAM files | Germline versus somatic calling strategy | Filtered VCF file with quality scores | | Database construction | VCF file plus reference proteome and transcript annotation | Include sample-specific variants only or supplement with public variant databases | Variant protein FASTA file with traceable headers | | Database search | Tandem mass spectra plus combined reference and variant FASTA | Combined single search versus separate parallel searches | Peptide-spectrum matches with scores | | FDR estimation | Search results from target and decoy databases | Variant-specific FDR threshold versus global FDR | Filtered peptide list at defined confidence | | Validation | Candidate variant peptides plus genomic variant list | Orthogonal genomic confirmation and spectral inspection | Validated variant peptide list | ## Generating a Variant Protein Database The first step in detecting variant peptides is to build a protein sequence database that includes both the reference proteins and the variant proteins expected in the sample. This database is then used for the peptide search. ### Source Genomic Data The starting point is genomic or transcriptomic data from the same biological sample that was analyzed by mass spectrometry. Ideally, the DNA or RNA sequencing data come from the same tissue or cell population. This pairing ensures that the variants included in the database are actually present in the sample being analyzed by proteomics. Several public resources provide access to genomic variation data. The National Center for Biotechnology Information (NCBI) maintains databases of genetic variation, reference genomes, and associated analysis tools. Researchers can use these resources to retrieve known variants for a species or to deposit and query their own variant calls. The European Bioinformatics Institute (EMBL-EBI) offers training materials and data resources for bioinformatics analysis, including practical guidance on working with genomic and proteomic data. For species with well-characterized variation, public variant databases such as dbSNP provide a starting point. For individual samples, variant calling from whole-genome, whole-exome, or RNA sequencing data is required. RNA sequencing has the advantage that it provides evidence of which variants are actually transcribed in the sample. ### Call Variants from Sequencing Data If you have raw sequencing data for your sample, you must first generate a list of variants. This involves aligning reads to the reference genome, calling variants, and filtering the calls for quality. Standard tools for this purpose include the Genome Analysis Toolkit for germline and somatic variant calling and freebayes for variant calling. The specific tool choice depends on your sequencing type and study design. The Galaxy Training Network provides accessible tutorials for variant calling and related analysis workflows. These tutorials are useful for researchers who want to learn the steps in a reproducible environment. The Carpentries offers foundational lessons in shell scripting, data management, and programming that are helpful for automating and documenting the variant calling process. For somatic mutation analysis in cancer samples, tumor-normal paired calling is recommended. This approach distinguishes somatic mutations from germline polymorphisms. For germline analysis in population studies, single-sample or cohort-based calling is appropriate. ### Translate Variants into Protein Sequences Once you have a list of genomic variants, you need to translate them into protein sequence changes. This step requires knowing the genomic coordinates of each variant, the gene in which it falls, the transcript isoform to which it applies, and the codon context. Several approaches exist for this translation step. One approach is to use a dedicated tool that accepts a variant file and a reference proteome and outputs a variant protein database. Another approach is to write a custom script that parses the variant file, identifies the affected codon or codons, applies the nucleotide change to the reference transcript sequence, and translates the modified transcript into protein. The Bioconductor project provides packages for genomic annotation and sequence manipulation that can be used to build custom variant translation workflows. Bioconductor packages are designed for reproducible genomic analysis and are distributed with documentation and examples. Researchers comfortable with R can use these packages to construct a variant database that is tailored to their specific sample and species. A key design decision is whether to include all possible variant peptides or only those with evidence in the sample. Including all known variants for a species can produce a very large database, which increases search time and can increase false discovery rates. Including only variants called in your sample keeps the database smaller and more relevant. ### Database Format and Structure The output of the translation step is a FASTA file containing protein sequences. Each entry should have a header that clearly identifies the protein as a variant and provides traceability. A recommended header format includes the gene name, the variant identifier or genomic coordinate, the amino acid change, and the source transcript. For example: ``` >TP53|chr17:7578406C>T|p.R248Q|ENST00000269305 ``` This header allows you to trace any identified variant peptide back to its genomic origin. Traceability is essential for validation and for reporting. The variant database should include the reference protein sequences as well. Some workflows append variant sequences to the reference database. Others create a separate variant-only database and search it in parallel. The choice affects downstream false discovery rate estimation, as discussed below. ## Search Workflow Design The variant database is used in the same search engines that are used for standard proteomics. Popular search engines include Sequest, Mascot, X!Tandem, and MyriMatch. A published workflow for variant peptide detection demonstrated compatibility with all four of these search engines. This compatibility means that researchers can integrate variant searching into their existing proteomics pipeline without changing search software. ### Search the Variant Database The search is performed in the same manner as a standard search. Tandem mass spectra are compared against the protein database, and peptide-spectrum matches are scored. The search parameters, including precursor mass tolerance, fragment mass tolerance, enzyme specificity, and fixed and variable modifications, should be identical to those used for the reference database search. One important consideration is that the search space is larger when variant sequences are included. This increases the number of candidate peptides and therefore increases the number of false positives at any given score threshold. False discovery rate control must account for this expanded search space. ### Combine or Separate Searches There are two common strategies for searching variant and reference databases. In the first strategy, the variant sequences are appended to the reference database, and a single search is performed against the combined database. In the second strategy, the reference and variant databases are searched separately, and the results are combined after the fact. The combined database approach is simpler and ensures that every spectrum is searched against both reference and variant sequences in a single pass. The separate search approach allows you to compare the results directly and to see how many spectra are uniquely identified by the variant database. A published workflow for variant peptide detection used a combined database approach and addressed the high risk of false positive variant identifications with a modified false discovery rate estimation method. This modified method is important because standard target-decoy approaches may not adequately control errors when the database contains many similar sequences. ### False Discovery Rate Control False discovery rate (FDR) estimation in proteomics typically uses a target-decoy strategy. A decoy database is generated by reversing or shuffling the target sequences, the search is performed against the concatenated target and decoy database, and the FDR is estimated from the number of decoy hits above the score threshold. For variant databases, this standard approach has a complication. Variant sequences are highly similar to their reference counterparts, often differing by a single amino acid. This similarity means that a decoy database generated from the variant sequences may not provide an accurate estimate of the false discovery rate for variant peptides. The published workflow mentioned above addressed this by modifying the FDR estimation method. The specific modification involved treating variant and reference identifications separately when estimating FDR. This separation prevents the large number of reference identifications from masking errors in the smaller set of variant identifications. When you design your own workflow, you should consider how to estimate FDR for variant peptides specifically. A conservative approach is to report variant peptides only if they pass a stricter FDR threshold than the overall dataset, such as 1% variant-specific FDR instead of 1% global FDR. ## Validation of Variant Peptide Identifications Variant peptide identifications require more scrutiny than standard identifications. The reason is that the search space is larger and the prior probability of any specific variant peptide being present is low. A spectrum that matches a variant peptide with a modest score may be a false match to a sequence that is similar to the true peptide. ### Orthogonal Validation with Genomic Data The strongest validation for a variant peptide is confirmation that the corresponding variant is present in the DNA or RNA of the same sample. If you called variants from sequencing data, you can check whether each candidate variant peptide corresponds to a variant call in your variant file. In the colorectal cancer study described above, the workflow detected 81 variant peptides in three cell lines. The researchers randomly selected 26 of these variants for genomic sequencing and confirmed 23 of them. This confirmation rate of approximately 88% demonstrates that variant peptide detection can be accurate, but it also shows that not every candidate survives validation. For your own workflow, you should establish a validation step that cross-references each candidate variant peptide against the genomic variant list. This step can be automated with a script that parses the peptide identification results and the variant file and reports which candidates have genomic support. ### Spectral Quality Assessment Variant peptides should have high-quality spectra. The fragment ion coverage should be complete enough to localize the variant site. A peptide that differs from the reference by one amino acid should show fragment ions that distinguish the two sequences. If the spectrum does not clearly support the variant residue at the specific position, the identification should be treated with caution. Manual inspection of spectra is recommended for the final set of candidate variant peptides. This inspection should confirm that the assigned fragment ions are consistent with the variant sequence and that the variant site is supported by fragment ions on both sides. ### Replicate Consistency If your experiment includes technical or biological replicates, a variant peptide that is detected in multiple replicates is more credible than one detected in a single replicate. Replicate consistency is not definitive proof, because a systematic error could affect all replicates, but it is a useful filter. ### Targeted Validation with Selected Reaction Monitoring For high-value candidate variant peptides, targeted mass spectrometry validation is appropriate. Selected reaction monitoring (SRM) can be used to detect a specific peptide with high sensitivity and specificity. A published study demonstrated the use of SRM to detect mutant p53 peptides in cancer cell lines and tumor samples. The study detected two mutant p53 peptides in two cancer cell lines as expected from their DNA sequences. In 18 individual tumor samples with TP53 mutations, SRM detected mutations at the protein level in two cytosolic extracts and whole tumor lysates from the same patients. This study illustrates both the potential and the limitations of targeted validation. SRM can confirm the presence of a specific variant peptide, but it requires prior knowledge of the peptide sequence and the development of a targeted assay. For discovery purposes, shotgun proteomics with a variant database is the appropriate first step. ## Incorporating Splice Variants Alternative splicing generates protein isoforms that are not represented in the reference proteome. Peptides that span novel exon-exon junctions are invisible to standard searches. Including these junction peptides in the search database requires a different approach than SNV translation. ### Generating Junction Peptide Databases The approach is to use RNA sequencing data to identify exon-exon junctions that are present in the sample. The sequences spanning these junctions are then translated in all three reading frames and added to the search database. The large-scale analysis of alternative splicing across cancer types identified thousands of alternative splicing events per tumor that were not detectable in normal samples. This finding suggests that splice variant peptides could be a rich source of tumor-specific antigens. The same study confirmed approximately 1.7 neojunction-derived peptides per tumor sample in available proteomics data, which is a modest but real detection rate. Tools for generating junction peptide databases include custom scripts that parse RNA sequencing alignments and extract junction sequences. The Galaxy Training Network provides tutorials for RNA sequencing analysis that can be adapted for this purpose. The Bioconductor project offers packages for transcript annotation and sequence extraction that can be used to build junction databases. ### Combining SNV and Splice Variant Databases A comprehensive variant database can include both SNV-derived and splice junction-derived sequences. The database construction process is more complex, but the search and validation steps are the same. One practical consideration is database size. Including all possible junction peptides from a transcriptome can produce a very large database. Filtering junctions by expression level or by the number of supporting reads can reduce the database to a manageable size. ## Practical Implementation Steps The following steps outline a complete workflow for incorporating genomic variants into proteomics searches. These steps assume that you have mass spectrometry data and either genomic variant calls or raw sequencing data for the same sample. ### Step 1: Assemble Your Inputs Collect the following inputs before you begin: - Tandem mass spectrometry data in a standard format such as MGF or mzML - A reference proteome FASTA file for your species - A variant call file (VCF) for your sample, or raw sequencing data to generate one - A transcript annotation file that maps genomic coordinates to transcripts and protein sequences If you do not have a variant call file, you must generate one from sequencing data. The Galaxy Training Network provides tutorials for this process. The Carpentries lessons provide foundational skills in shell and data management that are useful for handling large sequencing files. ### Step 2: Generate the Variant Protein Database Use a variant translation tool or a custom script to convert your variant call file into a variant protein FASTA file. The Bioconductor project provides packages that can perform this translation in a reproducible manner. For each variant, the translation script should: 1. Identify the affected transcript and codon position 2. Apply the nucleotide change to the reference transcript sequence 3. Translate the modified transcript into protein 4. Extract the variant protein sequence and write it to the FASTA file with a descriptive header For indels, the script must handle the reading frame shift correctly. For frameshift variants, the protein sequence downstream of the variant site will be entirely novel and should be included in the database. ### Step 3: Combine the Databases Append the variant protein sequences to the reference proteome FASTA file. Alternatively, keep them separate and search the two databases independently. The combined approach is simpler for downstream analysis. ### Step 4: Perform the Database Search Search your mass spectrometry data against the combined database using your preferred search engine. Use the same search parameters that you would use for a standard search. Record the search settings in your laboratory notebook or analysis documentation. ### Step 5: Estimate False Discovery Rate Apply FDR estimation with attention to the variant identifications. If your search engine or downstream tool supports separate FDR estimation for variant and reference peptides, use that option. Otherwise, apply a stricter threshold to variant peptides. ### Step 6: Filter and Validate Filter the search results to obtain a list of candidate variant peptides. Cross-reference each candidate against your genomic variant list. Manually inspect the spectra for the highest-confidence candidates. If replicates are available, check for replicate consistency. ### Step 7: Report Report the number of variant peptides identified, the validation status of each, and the FDR thresholds applied. Include the database construction parameters and the search parameters so that the analysis is reproducible. ## Records and Measurements Documentation is essential for a variant peptide detection workflow. The following records should be maintained for each analysis: ### Variant Call Records - The sequencing data source and alignment parameters - The variant calling tool and version - The filtering thresholds applied to variant calls - The number of variants called and the number passing filters ### Database Construction Records - The reference proteome version and source - The transcript annotation version and source - The variant translation tool or script version - The number of variant protein sequences added to the database - The final database size and composition ### Search Records - The search engine and version - The search parameters, including tolerances and modifications - The FDR estimation method and thresholds - The number of peptide-spectrum matches at each threshold ### Validation Records - The number of candidate variant peptides - The number with genomic support - The number confirmed by manual spectral inspection - The number confirmed by targeted mass spectrometry, if applicable These records allow you to reproduce the analysis and to compare results across experiments. They also provide the documentation needed for publication and for audit by collaborators or reviewers. ## Common Failure Patterns Several recurring problems arise in variant peptide detection. Recognizing these patterns can help you troubleshoot your workflow. ### Failure Pattern 1: No Variant Peptides Detected If your search returns no variant peptides, the possible causes include: - The variant database was not constructed correctly. Check that the variant translation step produced protein sequences and that these sequences are in the correct reading frame. - The variants are not expressed at the protein level. A variant may be present in the DNA but not in the protein because of nonsense-mediated decay, low expression, or protein degradation. - The search parameters are too strict. Variant peptides may have different mass spectrometry properties than reference peptides, particularly if the variant affects a cleavage site or a post-translational modification site. - The sample does not actually contain the variants. Confirm that the variant calls came from the same sample and that the sequencing depth was sufficient. The multi-tissue study that detected only 238 of 9,848 amino acid variants at the protein level illustrates that low detection rates are common. The authors attributed this to the challenges of proteogenomics and the need for better computational methods. If your detection rate is low, this may reflect biological reality instead of a workflow error. ### Failure Pattern 2: Too Many Variant Peptides Detected If your search returns an implausibly large number of variant peptides, the likely cause is inadequate FDR control. The expanded search space increases the number of false positives. Review your FDR estimation method and consider applying a stricter threshold to variant identifications. Another possible cause is that the variant database contains sequences that are not actually present in the sample. If you included all known variants for a species instead of only variants called in your sample, the database will contain many sequences that cannot be present. These sequences increase the search space and the false positive rate. ### Failure Pattern 3: Variant Peptides Not Reproducible If variant peptides detected in one run are not detected in a replicate, the possible causes include: - The variant peptide is present at low abundance and is detected stochastically - The variant peptide is a false positive in the first run - The sample differs between runs, for example due to tissue heterogeneity Replicate consistency is a useful filter, but it is not definitive. A variant peptide that is detected in only one of several replicates should be treated with caution. ### Failure Pattern 4: Genomic and Proteomic Data Disagree If a variant peptide is detected but the corresponding variant is not present in the genomic data, the possible causes include: - The variant peptide is a false positive - The genomic data came from a different sample or cell population - The variant is present at low allele frequency and was not called by the variant caller - The peptide sequence matches a different gene or a paralog This disagreement is a strong signal to reject the variant peptide identification unless there is a clear explanation. ## Limitations of Variant Peptide Detection Variant peptide detection has inherent limitations that should be understood before interpreting results. ### Detection Sensitivity The multi-tissue study that detected only 238 of 9,848 amino acid variants at the protein level provides a realistic estimate of detection sensitivity. The majority of amino acid variants present in the genome are not detected at the protein level. This is partly because many variants are in genes that are not expressed in the sampled tissue, and partly because mass spectrometry does not detect every peptide in a complex sample. ### Database Completeness The variant database is only as complete as the variant calls that generated it. Variants that are not called by the variant caller will not be in the database and cannot be detected. Low sequencing depth, alignment errors, and variant caller limitations all contribute to incomplete variant databases. ### False Discovery Control The expanded search space of a variant database increases the false discovery rate. Standard FDR methods may not adequately control errors for variant peptides because of the high sequence similarity between variant and reference proteins. Conservative thresholds and orthogonal validation are required. ### Biological Interpretation Detecting a variant peptide confirms that the variant is expressed at the protein level. It does not confirm that the variant has a functional effect. Many amino acid substitutions are neutral. Functional interpretation requires additional evidence, such as structural analysis, biochemical assays, or clinical correlation. ## Safety and Regulatory Context Variant peptide detection is a research methodology. It is not a clinical diagnostic test unless it has been validated and approved for that purpose. Researchers who use variant peptide detection for clinical samples should be aware of the regulatory requirements for laboratory-developed tests in their jurisdiction. The study that detected mutant p53 peptides in clinical samples noted the great potential of proteogenomics for direct identification of cancer-associated mutations in clinical samples, while also discussing current limitations and future perspectives. This balanced view is appropriate. The technology is promising, but it is not yet a routine clinical tool. For researchers working with human samples, institutional review board approval and informed consent requirements apply to the use of genomic and proteomic data. For researchers working with animal samples, institutional animal care and use committee approval may be required. These requirements are separate from the technical aspects of the workflow. ## Professional Escalation Criteria Variant peptide detection results should be escalated to a supervisor, collaborator, or specialist in the following situations: ### Unexpected Variant Patterns If you detect a large number of variant peptides in a sample where few were expected, or if the variant peptides cluster in a particular gene or pathway, this may indicate a sample quality issue, a database construction error, or a biologically significant finding. Escalate to a supervisor or collaborator to determine which interpretation is correct. ### Disagreement Between Genomic and Proteomic Data If your proteomics data consistently disagree with your genomic data, escalate to the bioinformatics specialist who performed the variant calling. The disagreement may indicate a systematic error in either the genomic or proteomic analysis. ### Potential Clinical Significance If you detect a variant peptide that corresponds to a known disease-associated mutation, and the sample is from a patient, escalate to the clinical team. The finding may have implications for patient care, but it must be confirmed by a validated clinical test before any clinical action is taken. ### Reproducibility Concerns If variant peptide identifications are not reproducible across replicates or across laboratories, escalate to the team to review the workflow. The lack of reproducibility may indicate a technical problem that affects the validity of the results. ## Frequently Asked Questions ### What is the difference between a variant peptide and a reference peptide? A variant peptide has an amino acid sequence that differs from the reference sequence for the species. The difference arises from a genomic variant such as a single nucleotide variant or an indel. A reference peptide matches the canonical protein sequence exactly. Variant peptides are not present in standard reference protein databases and require a variant-aware search workflow for detection. ### Why are variant peptides missed by standard proteomics searches? Standard proteomics searches compare spectra against reference protein databases that contain only canonical protein sequences. If a sample carries a genomic variant that changes the protein sequence, the peptide containing that change will not match any reference sequence. The search engine will either fail to assign the spectrum or assign it to an incorrect peptide. Including variant sequences in the search database is necessary to detect these peptides. ### What genomic data do I need to build a variant protein database? You need a list of variants for the same sample that was analyzed by mass spectrometry. This list can come from whole-genome, whole-exome, or RNA sequencing data. The variant list should include the genomic coordinates, the reference and alternate alleles, and the gene or transcript affected. Public variant databases can supplement sample-specific calls, but sample-specific data are essential for accurate detection. ### How do I translate genomic variants into protein sequence changes? Translation requires knowing the affected transcript, the codon position, and the nucleotide change. A dedicated tool or custom script can apply the nucleotide change to the reference transcript sequence and translate the modified transcript into protein. The Bioconductor project provides packages for this purpose. The output is a FASTA file containing variant protein sequences with descriptive headers. ### How do I control false discoveries in variant peptide searches? False discovery rate control is more challenging for variant searches because the search space is larger and variant sequences are highly similar to reference sequences. A published workflow used a modified FDR estimation method that treats variant and reference identifications separately. A conservative approach is to apply a stricter FDR threshold to variant identifications and to validate candidates with orthogonal evidence. ### What validation is required for variant peptide identifications? The strongest validation is confirmation that the corresponding variant is present in the genomic data from the same sample. Manual inspection of spectra is recommended to confirm that the variant site is supported by fragment ions. Replicate consistency provides additional confidence. For high-value candidates, targeted mass spectrometry with selected reaction monitoring can confirm the presence of the specific variant peptide. ### Can I detect splice variant peptides with the same workflow? Splice variant peptides require a different database construction approach. Instead of translating genomic variants, you identify exon-exon junctions from RNA sequencing data and translate the junction sequences in all reading frames. The search and validation steps are the same as for SNV-derived variant peptides. A large-scale cancer study identified approximately 930 novel exon-exon junctions per tumor, but confirmed only about 1.7 neojunction-derived peptides per tumor in available proteomics data. ### What should I do if I detect a variant peptide with potential clinical significance? A variant peptide with potential clinical significance should be confirmed by a validated clinical test before any clinical action is taken. The proteomics finding should be escalated to the clinical team and documented. The regulatory requirements for laboratory-developed tests apply if the assay is used for clinical decision-making. The research finding is valuable for hypothesis generation, but it is not a substitute for a validated diagnostic test. ## Related Bioinformatics Guides - [Genomic Data Analysis Tools: A Comparative Guide for Researchers](/knowledge/bioinformatics/genomic-data-analysis-tools-a-comparative-guide-for-researchers) - [Spatial Proteomics vs. Single-Cell Proteomics: Choosing the Right Approach](/knowledge/bioinformatics/spatial-proteomics-vs-single-cell-proteomics-choosing-the-right-approach) - [Genomic Data Infrastructure: Building and Managing Large-Scale Genomic Databases](/knowledge/bioinformatics/genomic-data-infrastructure-building-and-managing-large-scale-genomic-databases) - [Genomic Data Repositories: Navigating Public Databases for Research](/knowledge/bioinformatics/genomic-data-repositories-navigating-public-databases-for-research) - [Single-Cell Sequencing Databases: Resources for Data Sharing and Exploration](/knowledge/bioinformatics/single-cell-sequencing-databases-resources-for-data-sharing-and-exploration) ## 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. - [Comprehensive Analysis of Alternative Splicing Across Tumors from 8,705 Patients.](https://pubmed.ncbi.nlm.nih.gov/30078747). Cancer cell, 2018. - [A deep proteome and transcriptome abundance atlas of 29 healthy human tissues.](https://pubmed.ncbi.nlm.nih.gov/30777892). Molecular systems biology, 2019. - [GDF15 linked to maternal risk of nausea and vomiting during pregnancy.](https://pubmed.ncbi.nlm.nih.gov/38092039). Nature, 2024. - [Variant peptide detection utilizing mass spectrometry: laying the foundations for proteogenomic identification and validation.](https://pubmed.ncbi.nlm.nih.gov/28157690). Clinical chemistry and laboratory medicine, 2017. - [A bioinformatics workflow for variant peptide detection in shotgun proteomics.](https://pubmed.ncbi.nlm.nih.gov/21389108). Molecular & cellular proteomics : MCP, 2011. > This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.