1. Reproducibility standards
Ask whether the CRO version-pins software, logs non-default parameters, and delivers runnable scripts with documentation, not PDF reports alone. Sandve et al.'s [8] ten rules are a useful reference, particularly the rule about archiving exact software versions. Request a sample `environment.yml` or `requirements.txt` from a completed project.
2. Code and data ownership
Confirm in writing that your institution retains full ownership of custom code, processed outputs, and analysis artifacts. Clarify whether the CRO retains reusable pipeline IP and what that means for your licensing. Unresolved IP questions tend to surface at publication, when there is less room to renegotiate.
3. Data security
Require encrypted transfer (SFTP, AWS S3 with SSE-KMS, or equivalent), isolated compute environments per client, and a project-specific NDA before any FASTQ files leave your institution. Ask where data is stored, retention period after project close, and deletion certification. Human-subject or clinical data may need BAA or GDPR-compliant processing—verify the CRO has handled comparable data classes.
4. Publication track record
Ask for peer-reviewed papers where team members contributed analysis or co-authored omics work comparable to yours in modality, organism, and sample size. A newer CRO may have no client projects under its own name yet; team publication history is often the more useful signal. Track record is not a guarantee, but no citable omics work in your domain is worth asking about.
5. Communication cadence
Define a dedicated scientific contact and a standing meeting schedule: weekly during active analysis, biweekly during reporting. Make sure you can reach a scientist who understands your biological question, not only a project manager.
6. Milestone pricing and scope
Fixed-price milestones tied to deliverables—QC report, primary analysis, figure package, Methods draft—are easier to budget than open-ended hourly billing. Each milestone should list acceptance criteria. Ask what triggers a change order and how long revisions typically take.
7. Reviewer support
Clarify whether post-submission reviewer questions are included or billed separately. Requests for code, parameter logs, or re-analysis of subsets are common; your SOW should say who responds and within what timeframe.
8. Modality and pipeline expertise
Match the CRO to your data type. Bulk RNA-seq, single-cell, spatial transcriptomics, WGS, proteomics, and metagenomics each require different QC norms and reference builds. Ask which pipelines they run routinely, whether they use community frameworks, and how they handle novel or poorly annotated genomes.
9. Turnaround realism
Request typical timelines for projects of your scale, including queue time, not best-case estimates. Ask whether expedited delivery costs extra and whether rush schedules affect QA depth. A 60-sample RNA-seq study does not produce manuscript-ready output in 48 hours.