Proteomics analysis services: reproducible quantification from DDA, DIA, single-cell MS, and Olink assays

Clark et al. [16]: 820 differential proteins in 103 tumors vs 80 normal adjacent tissues, including oxidative phosphorylation changes visible only at the protein level.

Specht et al. [17]: 3,042 proteins across 1,490 single cells (SCoPE2), revealing a continuous proteome gradient.

Wigren et al. [18]: decreased stem cell factor before events, measured by proximity extension assay.

Clark et al. [16]: phosphoproteomics linked kinase modules to a genomic-instability ccRCC subtype.

Bäckryd et al. (2017): distinct CSF vs plasma inflammatory profiles in fibromyalgia using a multiplex immunoassay panel.

MNAR and MCAR missingness can inflate false positives if they are imputed without checking the pattern [3]. Pepkio profiles missingness and documents the imputation or complete-case strategy before testing.

Column changes, instrument drift, and prep dates can introduce non-biological variance [4]. Batch diagnostics come before modeling. Batch-in-design or proBatch correction is used only when batch is not confounded with condition.

Translation, degradation, and localization can decouple transcript and protein abundance [16]. Proteogenomic correlation reports can be added at kickoff when matched RNA is available.

Thermo `.raw` and Bruker `.d` load in MaxQuant or FragPipe; SCIEX `.wiff` converts to mzML before search when needed.

Aggressive grouping can collapse isoforms. Grouping rules are documented in the Methods draft. Strict and expanded protein lists can be provided for reviewer sensitivity analyses.

Provide raw MS files (`.raw`, `.d`, `.wiff`, or mzML), a sample metadata table with condition and batch columns, and the organism plus UniProt database preference. Olink projects require NPX export files and the panel lot report. Pre-processed protein matrices are accepted when upstream search parameters are documented. Custom input formats and non-standard metadata structures are supported when scoped at kickoff.

Standard label-free or DIA projects (10-30 samples, one contrast) typically complete in 2-4 weeks. TMT, single-cell MS, multi-panel Olink, or phosphoproteomics extensions may take 4-8 weeks. Milestone check-ins occur after QC and before delivery. Exact timelines are confirmed at kickoff.

You receive protein and peptide matrices, differential-abundance tables, volcano plots and PCA figures (PDF/SVG), an HTML QC report, commented R or Python scripts with conda or `renv` lock files, a README, and a Methods draft. Optional GitHub or Zenodo archival is available on request.

Pepkio processes Thermo `.raw` and Bruker `.d` files in MaxQuant 2.8.0.0 or FragPipe 22+ with MSFragger. SCIEX `.wiff` files are converted to mzML before search when needed, and mzML or mzXML files are also accepted. The Methods draft records the search engine, database version, and FDR thresholds applied.

Pepkio documents identification rates, missing-value fractions, and outlier samples in the QC report before differential testing. High missingness triggers a review of whether MNAR-aware methods (proDA), Perseus imputation, or complete-case analysis is a better fit [3, 11]. Re-analysis after sample re-measurement is discussed when QC may undermine the study question.

You retain full ownership of all code and results. Pepkio delivers commented R and Python scripts with conda or `renv` lock files listing exact package versions. Scripts are organized by pipeline stage with a README describing execution from raw inputs to final tables.

Checkpoint reviews occur after raw-data QC, after normalization and batch diagnostics, and before final delivery. You can review contrast definitions, covariate inclusion, imputation choices, and FDR thresholds within agreed scope. A dedicated scientific contact leads the project and is available for scheduled calls.

Post-delivery support covers clarification of methods, QC thresholds, and minor figure or table revisions within agreed scope (typically ≤20% of deliverables). Substantial new analyses, such as additional contrasts, re-imputation with alternate methods, or proteogenomic extensions, are scoped as separate milestones.

Pepkio imports Olink NPX Signature exports, applies per-sample QC flagging (incubation and detection controls), and performs between-plate normalization using bridge samples or intensity normalization as appropriate for the study design [5]. Differential testing uses limma or MSstats, with the normalization method documented in the Methods draft.

Bespoke analyses, including custom contrasts, proteogenomic integration, phosphoproteomics extensions, ML classifiers, or non-standard deliverables, are scoped at kickoff after a feasibility review.