Bioinformatics analysis service
Bulk, single-cell, spatial, and long-read RNA-seq from FASTQs or vendor exports through QC to annotated tables, figures, scripts you keep, and a Methods draft.
| Category | Details |
|---|---|
| Data types supported | Bulk RNA-seq (Illumina or MGI DNBSEQ FASTQs or count matrices); 10x Chromium scRNA-seq; 10x Visium/Xenium spatial transcriptomics; PacBio HiFi and Oxford Nanopore long-read RNA-seq |
| Reference builds or standards used | Human GRCh38 (GENCODE v44 / Ensembl 110); mouse GRCm39 (GENCODE vM33 / Ensembl 110); ENCODE RNA-seq standards for bulk depth and replicate guidance [3] |
| Primary tools (with versions) | STAR 2.7.11b; DESeq2 1.52.0; Cell Ranger 10.0.0; Scanpy 1.12.1; Seurat 5.2.1; Space Ranger 3.1.3; Squidpy 1.8.1; IsoQuant 3.13.0; SQANTI3 6.0.1; clusterProfiler 4.20.0 |
| Typical turnaround range | 2-4 weeks (bulk); 3-5 weeks (scRNA-seq); 4-6 weeks (spatial or long-read); 5-10 weeks (multi-cohort integration or HD spatial), confirmed at kickoff |
| Deliverable formats | .csv/.tsv count matrices; .rds/h5ad objects; .gtf isoform catalogs; PDF/SVG figures; HTML QC reports; commented R/Python scripts; Methods draft |
| Regulatory/reproducibility standards followed | ENCODE RNA-seq guidelines; MINSEQE-aligned sample metadata and reporting; version-pinned software with sessionInfo() or conda lock files; private Git or Zenodo archival on request |
| Custom / bespoke analysis | Non-standard inputs (pre-aligned BAM, vendor matrices), output formats, methods, and analyses beyond standard workflows, for example cross-modality integration, fusion calling, trajectory inference, or client-specified contrasts, scoped at kickoff |
Transcriptomics is the genome-wide measurement of RNA, including messenger RNA (mRNA), long non-coding RNA, and other transcript classes. It is used to quantify which genes are expressed, at what level, and in which cells or tissue regions. Unlike targeted assays such as qPCR, RNA sequencing (RNA-seq) can capture novel splice junctions and unannotated transcripts when depth permits [1]. Typical questions include which transcripts differ between conditions, cell types, or spatial domains, and which pathways those changes implicate. GTEx v8 profiled 17,382 RNA-seq samples across 52 tissues and two cell lines from 948 donors [2].
Published examples of questions transcriptomics can address:
Sun et al. [13] analyzed TCGA bulk RNA-seq from 103 primary and 368 metastatic melanoma samples, identifying 246 differentially expressed lncRNAs and 856 mRNAs between stages.
Liu et al. [9] profiled 47 NSCLC biopsies from 36 patients before and after PD-1 therapy, showing precursor exhausted T cells accumulate in responders through clonal revival rather than reinvigoration of terminally exhausted cells.
Arora et al. [14] applied spatial transcriptomics to oral squamous cell carcinoma, identifying conserved tumor core and leading-edge domains linked to survival and therapy response.
Amarasinghe et al. [15] reviewed long-read RNA-seq applications showing that full-length reads resolve isoform structures that short-read fragments cannot reconstruct reliably.
GTEx v8 linked cis-expression quantitative trait loci (eQTLs) across 52 tissues and two cell lines, enabling tissue-context interpretation of regulatory variants [2].
This category includes bulk RNA-seq, single-cell RNA-seq, spatial transcriptomics, and long-read RNA-seq. Each spoke page lists inputs, tools, and deliverables.
| Service | Description | Primary tools |
|---|---|---|
| Bulk RNA-seq | Pooled-tissue differential expression from FASTQs or count matrices to DESeq2 contrasts and pathway enrichment | STAR 2.7.11b, featureCounts 2.0.8, DESeq2 1.52.0 |
| Single-Cell RNA-seq | Per-cell clustering, annotation, and differential expression from 10x Chromium FASTQs to annotated UMAPs | Cell Ranger 10.0.0, Scanpy 1.12.1, Seurat 5.2.1 |
| Spatial Transcriptomics | Tissue-coordinate expression, spatial domains, and cell-type deconvolution from Visium or Xenium data | Space Ranger 3.1.3, Squidpy 1.8.1, cell2location 0.1.5 |
| Long-Read RNA-seq | Full-length isoform catalogs and isoform-level differential expression from PacBio HiFi or ONT reads | minimap2 2.31, IsoQuant 3.13.0, SQANTI3 6.0.1 |
Each project includes processed data, figures, scripts, and a Methods draft. Summary slides are not the only output:
Non-standard contrasts, cross-modality integration, or client-specified figures are scoped at kickoff.
Confirm biological question, replicates, platform, and modality. Flag confounded batch or a modality mismatch before analysis [1, 4].
Output: signed scope with contrasts
Verify FASTQ integrity, chemistry, and sample IDs; check vendor outputs against annotation builds.
Tools used: FastQC 0.12.1
Output: sample_manifest.csv
Assess adapter contamination and library quality per sample; aggregate metrics for review [5].
Tools used: fastp 0.24.0; MultiQC 1.25.1; NanoPlot 1.42.0 (long-read)
Output: multiqc_report.html
Bulk: STAR 2.7.11b + featureCounts 2.0.8. scRNA-seq: Cell Ranger 10.0.0. Spatial: Space Ranger 3.1.3. Long-read: minimap2 2.31 + IsoQuant 3.13.0 [6, 7].
Output: count matrices and alignment summaries
Compute library sizes, detection rates, and read-distribution metrics. PCA is used to separate condition from batch when possible [8, 9].
Tools used: DESeq2 1.52.0; RSeQC 5.0.3 [10]
Output: sample_qc_summary.csv; PCA plots
Bulk: DESeq2 Wald tests with batch covariates [8]. scRNA-seq: clustering and cluster-wise DE. Spatial: Squidpy domains and pseudobulk statistics. Long-read: SQANTI3 filtering plus DESeq2 or DRIMSeq when replicates support isoform testing.
Output: deg_results.csv; annotated objects
GO/KEGG enrichment; cell-type annotation; cell2location deconvolution when a reference atlas exists [10, 11].
Tools used: clusterProfiler 4.20.0; SingleR 2.12.0
Output: enrichment and annotation tables
Export figures at publication resolution; package scripts, environment locks, README, and Methods draft.
Output: PDF/SVG figures; Git repository or file bundle
Transfer deliverables via agreed secure channels. Post-delivery support covers methods clarification and minor revisions; substantial new analyses are scoped separately.
Software versions are pinned at kickoff and cited in the Methods draft. Tools used across modalities include:
| Tool | Version | Role | Primary citation |
|---|---|---|---|
| STAR | 2.7.11b | Splice-aware bulk RNA-seq alignment | https://doi.org/10.1093/bioinformatics/bts635 |
| Subread/featureCounts | 2.0.8 | Gene-level read counting | https://doi.org/10.1093/bioinformatics/btt656 |
| DESeq2 | 1.52.0 | Bulk and isoform-level differential expression | https://doi.org/10.1186/s13059-014-0550-8 |
| Cell Ranger | 10.0.0 | 10x scRNA-seq alignment and UMI counting | 10x Genomics [12] |
| Scanpy | 1.12.1 | scRNA-seq and spatial object processing (Python) | https://doi.org/10.1186/s13059-017-1382-0 |
| Seurat | 5.2.1 | scRNA-seq and spatial object processing (R) | https://doi.org/10.1038/s41587-023-01767-y |
| Space Ranger | 3.1.3 | 10x Visium alignment and spot quantification | 10x Genomics [12] |
| Squidpy | 1.8.1 | Spatial statistics, neighborhood graphs, SVG detection | https://doi.org/10.1038/s41592-021-01358-2 |
| minimap2 | 2.31 | Long-read RNA-seq alignment | https://doi.org/10.1093/bioinformatics/bty191 |
| IsoQuant | 3.13.0 | Long-read transcript discovery and quantification | https://doi.org/10.1038/s41587-022-01565-y |
| SQANTI3 | 6.0.1 | Long-read isoform classification and filtering | https://doi.org/10.1038/s41592-024-02229-2 |
| clusterProfiler | 4.20.0 | GO/KEGG enrichment | https://doi.org/10.1089/omi.2011.0118 |
| MultiQC | 1.25.1 | Aggregated QC reporting | https://doi.org/10.1093/bioinformatics/btw354 |
Reference builds follow GENCODE v44 (human) and GENCODE vM33 (mouse) unless a project needs a custom annotation. For bulk depth, ENCODE recommends at least 30 million aligned reads per replicate [3]. Inferential designs typically need at least 3 biological replicates per condition when that is feasible [1].
Common issues include pipeline choice, batch confounding, incomplete reporting, QC imbalances, and modality mismatch. Pepkio uses version-pinned workflows and checkpoint reviews.
192 pipeline combinations on 18 samples yielded materially different quantification and DE results [16]. Pepkio selects version-pinned tool chains at kickoff and documents choices in the Methods draft.
Technical variation can mask biological signal [4, 17]. Pepkio includes batch in design formulas when it is not confounded with condition, and reviews PCA before modeling.
Only 25% of RNA-seq articles describe all essential computational steps [18]. Pepkio delivers scripts, parameter logs, and a Methods draft with exact versions.
35% of 40 clinical RNA-seq datasets had significant quality imbalances between groups [19]. Pepkio audits per-sample QC and flags asymmetry before testing.
Bulk RNA-seq averages across cell types, so heterogeneous tissues may need scRNA-seq or spatial resolution [1, 11]. Pepkio discusses modality at kickoff.
Provide FASTQs, vendor outputs (Cell Ranger, Space Ranger), or count matrices plus metadata listing condition, batch, and covariates. Pepkio confirms chemistry and reference build at kickoff. For bulk RNA-seq, ENCODE recommends at least 30 million aligned reads per replicate [3]. Custom formats are accepted when scoped in advance.
Bulk projects usually take 2-4 weeks; scRNA-seq 3-5 weeks; spatial or long-read 4-6 weeks; and multi-cohort or HD spatial 5-10 weeks. Exact timelines are confirmed at kickoff.
Deliverables include count matrices or annotated objects, DEG tables, PDF/SVG figures, an HTML QC report, commented scripts, and a Methods draft citing tool versions.
Yes. Supported inputs include Illumina and MGI DNBSEQ bulk data (T7, G400, G99 when scoped at kickoff), 10x Chromium, Visium/Xenium, and PacBio or ONT long-read data. Visium HD, BD Rhapsody, Parse, CosMx, and pre-built matrices are also supported when scoped at kickoff.
Sub-threshold libraries can still be analyzed, with limitations documented in the QC report [3]. Outliers are flagged before modeling. Re-sequencing is discussed when yield may undermine the study question. Quality imbalances between condition groups are reported [19].
Yes. You retain full ownership. Pepkio delivers commented R/Python scripts with environment lock files via private Git or agreed file transfer.
Yes. Checkpoint reviews follow QC and the sample audit, and another review happens before final delivery. You can review contrasts, covariates, and filtering thresholds within agreed scope.
Methods clarification and minor revisions within agreed scope (typically up to 20% of deliverables) are covered. Substantial new analyses are scoped and priced separately.
Bulk is often enough when tissue composition is stable. scRNA-seq is better for cell-type heterogeneity, spatial methods keep tissue architecture, and long-read data is better for isoforms [1, 4, 11, 15]. Pepkio can help choose at kickoff.
Yes, when scoped at kickoff: custom inputs, fusion calling, trajectory inference, cross-modality integration, alternative DE engines (edgeR, limma-voom), or client-specified outputs.
Deep-dive pages for specific transcriptomics methods and workflows.
Tell us:
We'll tell you:
Contact us to start with a free consultation. Need everyday bench calculators? Try our free lab tools.
