C105 / R0123

Challenge C105-CH2 · approved draft version 1

summary

R0123 is a strong technical match for C105-CH2: a bioinformatics/HPC researcher with registered evidence on cloud cost modelling for alignment and variant-calling pipelines, Snakemake scatter–gather throughput improvement with concordance checks, and a documented industry optimisation engagement, aligned with the company's unprofiled but suspected alignment/calling bottlenecks.

Claims: C1, C3, C4, C5, C6

technicalRationale

The evidence covers the two likely levers: (a) cost-aware execution on spot/preemptible instances with checkpointing, and (b) parallelism/scatter–gather restructuring of alignment and calling stages, both with explicit accuracy-concordance measurements rather than speed alone.

Claims: C3, C4, C6

Project GENOMFLO develops a cost-aware orchestration layer and containerised alignment/variant-calling components for laboratories with limited bioinformatics staff, which resembles the company's two-person bioinformatics setup.

Claims: C2, C6

cooperationScope

A time-boxed pilot (roughly 13 weeks) under contract research or consulting: a 4-week profiling and cost-model phase, a 4-week optimisation of alignment/calling parallelism and checkpointed spot execution, and a 4-week concordance validation and handover to the company's two bioinformatics staff.

Claims: C7, C8, C10

validation

Acceptance gates: per-rule profile before/after, compute cost per sample measured against provider billing, total batch wall time, and star-allele call concordance versus the current pipeline on retained ethics-approved samples, with clinical accuracy treated as a non-negotiable constraint.

Claims: C7, C8

Concordance methodology should follow the cited practice of comparing against the existing baseline with precision/recall style metrics, not just timing.

Claims: C4, C8

constraints

Reported cost and runtime gains derive from WGS/WES workloads and modelled spot markets; they are not guarantees for a targeted star-allele panel, and spot usage without checkpointing can increase effective cost.

Claims: C3, C4, C8

Collaboration preferences support contract research, consulting and joint projects; availability, fees, IP and data-processing agreements are not documented and must be agreed separately.

Claims: C10

unknowns

The company's actual per-step bottlenecks, instance types and billing detail are unmeasured, so the sub-3-EUR and under-90-minute targets remain untested hypotheses.

Claims: C6, C7

Identity of co-authors and co-participants on the cited records is partly unresolved in the source data; no conclusions are drawn about their roles.

Claims: C9

Claims, evidence and limitations

C1 (fact)

R0123's profile lists research fields Bioinformatics, High-Performance Computing, Genomics and Cloud Computing, with keywords including read alignment, variant calling, pipeline parallelisation and cloud cost optimisation.

Profile self-description, not verified against project deliverables.

Evidence: ev-54343b077a33e039b3a4a0b795a33ae0741442ddfdf57958b684a6accb6f84b8, ev-625978e55e235393f924c2e318d2719d442f8dbb4c303d42744eac77a938f3e5

[
  {
    "evidenceId": "ev-54343b077a33e039b3a4a0b795a33ae0741442ddfdf57958b684a6accb6f84b8",
    "fieldPointer": "/fields",
    "relation": "profile_field",
    "subjectId": "R0123",
    "objectId": "R0123"
  },
  {
    "evidenceId": "ev-625978e55e235393f924c2e318d2719d442f8dbb4c303d42744eac77a938f3e5",
    "fieldPointer": "/keywords",
    "relation": "profile_field",
    "subjectId": "R0123",
    "objectId": "R0123"
  }
]

C2 (fact)

R0123 is a participant in project SYN-PRJ-0254 (GENOMFLO), which develops a cloud-cost-aware orchestration layer that dynamically chooses between on-premises HPC and spot-instance cloud execution based on queue state, sample priority and real-time cloud pricing, plus containerised alignment and variant-calling workflow components.

Project summary describes planned pilot deployments at two regional clinical partners; no completed-deployment results are given in the evidence.

Evidence: ev-c7e8de2c805b759e633e2f6328735e9f087252e1884af5370d37272fa10162f5, ev-5f1f974c5f0df9b942354f61a09d05cf882becdcd97f0825ff9f33df178b195e

[
  {
    "evidenceId": "ev-c7e8de2c805b759e633e2f6328735e9f087252e1884af5370d37272fa10162f5",
    "fieldPointer": "/summary",
    "relation": "source_field",
    "subjectId": "SYN-PRJ-0254",
    "objectId": "SYN-PRJ-0254"
  },
  {
    "evidenceId": "ev-5f1f974c5f0df9b942354f61a09d05cf882becdcd97f0825ff9f33df178b195e",
    "fieldPointer": "/participants/0/researcher_id",
    "relation": "project_participant",
    "subjectId": "R0123",
    "objectId": "SYN-PRJ-0254"
  }
]

C3 (fact)

R0123 co-authored work building a cost model for WGS alignment and variant-calling pipelines that combines hourly spot price distributions from two cloud providers, mean time to interruption, and checkpoint overhead for BWA-MEM2 and GATK stages; naive spot use without checkpointing increased effective cost 18–34% in high-volatility markets, whereas a bidding-and-checkpoint strategy cut cost 41% on average versus on-demand in stable price periods.

Modelled on WGS/WES cohorts at 10×–30× coverage, not on targeted short-read star-allele panels; transfer to a 24-sample panel batch is an assumption.
Pricing sampled over 90 days from two providers; results are sensitive to market volatility.

Evidence: ev-17eb6d3e27a3376b2f2e3e4d515dba94c575628055a59c5a949a8afe31768b2f, ev-127c213f3b14f099f76885e8f4d49b21e3dc6379799d497ca809e517dd364441, ev-e2936af7872e400217df9e2d304f79c3990e5532e132eda75307dfdb3afc7aa8

[
  {
    "evidenceId": "ev-17eb6d3e27a3376b2f2e3e4d515dba94c575628055a59c5a949a8afe31768b2f",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01615",
    "objectId": "SYN-PUB-01615"
  },
  {
    "evidenceId": "ev-127c213f3b14f099f76885e8f4d49b21e3dc6379799d497ca809e517dd364441",
    "fieldPointer": "/authors/0/researcher_id",
    "relation": "author_of",
    "subjectId": "R0123",
    "objectId": "SYN-PUB-01615"
  },
  {
    "evidenceId": "ev-e2936af7872e400217df9e2d304f79c3990e5532e132eda75307dfdb3afc7aa8",
    "fieldPointer": "/keywords",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01615",
    "objectId": "SYN-PUB-01615"
  }
]

C4 (fact)

R0123 co-authored a study on Snakemake/SLURM orchestration for GATK HaplotypeCaller, where genomic-interval scatter–gather reduced mean per-sample runtime from 4.1 hours to 41 minutes on 32 cores, with SNP precision 99.91%/recall 99.87% and indel precision 99.34%/recall 99.21% against the baseline on GIAB variants.

WES cohort at mean 85× coverage; the accuracy comparison is concordance with the authors' own baseline, not an independent clinical validation.

Evidence: ev-386ebffa1089ce50395b10e140c2bf00d9432f00b34aeb4c55f9d96b8cbee020, ev-6d8d59a890ebaab690cdd83583375e0c20744530a7560717c74f0deddebb53fc

[
  {
    "evidenceId": "ev-386ebffa1089ce50395b10e140c2bf00d9432f00b34aeb4c55f9d96b8cbee020",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01614",
    "objectId": "SYN-PUB-01614"
  },
  {
    "evidenceId": "ev-6d8d59a890ebaab690cdd83583375e0c20744530a7560717c74f0deddebb53fc",
    "fieldPointer": "/authors/3/researcher_id",
    "relation": "author_of",
    "subjectId": "R0123",
    "objectId": "SYN-PUB-01614"
  }
]

C5 (fact)

R0123 has documented industry experience: a 2023 three-month consulting engagement with a Novi Sad health-tech startup auditing and optimising a cloud-based WES pipeline, reportedly reducing per-sample cloud spend by about 35%.

Self-reported profile field; no client reference, baseline figures or independent verification in the evidence.

Evidence: ev-226e5222e55fdfeb9a9aa56e0f49f3737b4529536d32a22e8158d3bfed47f3f4

[
  {
    "evidenceId": "ev-226e5222e55fdfeb9a9aa56e0f49f3737b4529536d32a22e8158d3bfed47f3f4",
    "fieldPointer": "/industry_experience",
    "relation": "profile_field",
    "subjectId": "R0123",
    "objectId": "R0123"
  }
]

C6 (inference)

Because the company's bottlenecks are unprofiled but suspected in alignment and star-allele calling, R0123's combination of alignment-parallelisation work, scatter–gather throughput evaluation and spot/checkpoint cost modelling maps onto the primary levers likely needed to move from 9 EUR/sample toward below 3 EUR/sample.

Neither the company nor R0123 has measured this specific pipeline; mapping is by analogy from other WGS/WES pipelines.
Explicitly a hypothesis until per-step profiling is done.

Evidence: ev-386ebffa1089ce50395b10e140c2bf00d9432f00b34aeb4c55f9d96b8cbee020, ev-17eb6d3e27a3376b2f2e3e4d515dba94c575628055a59c5a949a8afe31768b2f, ev-756362a0e60229f6e2fa04f25c74f98b4647f4985d053fae0527384acb43b07d, ev-10ea081c1d4ce5d21da7b366caba15ed185c103fa10706ccbc59178341f15d0c

[
  {
    "evidenceId": "ev-386ebffa1089ce50395b10e140c2bf00d9432f00b34aeb4c55f9d96b8cbee020",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01614",
    "objectId": "SYN-PUB-01614"
  },
  {
    "evidenceId": "ev-17eb6d3e27a3376b2f2e3e4d515dba94c575628055a59c5a949a8afe31768b2f",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01615",
    "objectId": "SYN-PUB-01615"
  },
  {
    "evidenceId": "ev-756362a0e60229f6e2fa04f25c74f98b4647f4985d053fae0527384acb43b07d",
    "fieldPointer": "/title",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01612",
    "objectId": "SYN-PUB-01612"
  },
  {
    "evidenceId": "ev-10ea081c1d4ce5d21da7b366caba15ed185c103fa10706ccbc59178341f15d0c",
    "fieldPointer": "/bio",
    "relation": "profile_field",
    "subjectId": "R0123",
    "objectId": "R0123"
  }
]

C7 (proposal)

A first, bounded pilot: instrument the existing Snakemake pipeline to produce per-rule wall-time, CPU-hour and I/O profiles on two 24-sample batches, then build a per-sample cost model calibrated to the company's actual provider pricing and instance families, before any code optimisation.

Requires access to the company's cloud billing data and pipeline configuration, which are not in the evidence.
Cost-model outputs depend on provider price volatility during the pilot window.

Evidence: ev-17eb6d3e27a3376b2f2e3e4d515dba94c575628055a59c5a949a8afe31768b2f, ev-386ebffa1089ce50395b10e140c2bf00d9432f00b34aeb4c55f9d96b8cbee020, ev-c7e8de2c805b759e633e2f6328735e9f087252e1884af5370d37272fa10162f5

[
  {
    "evidenceId": "ev-17eb6d3e27a3376b2f2e3e4d515dba94c575628055a59c5a949a8afe31768b2f",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01615",
    "objectId": "SYN-PUB-01615"
  },
  {
    "evidenceId": "ev-386ebffa1089ce50395b10e140c2bf00d9432f00b34aeb4c55f9d96b8cbee020",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01614",
    "objectId": "SYN-PUB-01614"
  },
  {
    "evidenceId": "ev-c7e8de2c805b759e633e2f6328735e9f087252e1884af5370d37272fa10162f5",
    "fieldPointer": "/summary",
    "relation": "source_field",
    "subjectId": "SYN-PRJ-0254",
    "objectId": "SYN-PRJ-0254"
  }
]

C8 (proposal)

Targeted optimisation phase using interval scatter–gather and sample-level parallelism for the alignment and calling stages, plus checkpointed spot/preemptible execution guided by the cost model, with concordance against the current calls measured on retained clinical samples as the accuracy gate for the under-90-minute and sub-3-EUR goals.

Runtime and cost reductions reported in the cited studies were obtained on WGS/WES workloads, not on a targeted star-allele panel; gains may be smaller.
Effort assumes a 13-week pilot at the modest pace the proposer assigns, i.e. a 4-week plan, monitoring and a 4-week optimisation phase, with a 4-week accuracy-validation and handover period.
Spot-instance strategies carry interruption risk and require checkpointing to avoid cost regressions.

Evidence: ev-386ebffa1089ce50395b10e140c2bf00d9432f00b34aeb4c55f9d96b8cbee020, ev-17eb6d3e27a3376b2f2e3e4d515dba94c575628055a59c5a949a8afe31768b2f, ev-b883d6b162209ccffcea999a4f248970d03cc2988f8946f05ca2ee73c3871e94

[
  {
    "evidenceId": "ev-386ebffa1089ce50395b10e140c2bf00d9432f00b34aeb4c55f9d96b8cbee020",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01614",
    "objectId": "SYN-PUB-01614"
  },
  {
    "evidenceId": "ev-17eb6d3e27a3376b2f2e3e4d515dba94c575628055a59c5a949a8afe31768b2f",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01615",
    "objectId": "SYN-PUB-01615"
  },
  {
    "evidenceId": "ev-b883d6b162209ccffcea999a4f248970d03cc2988f8946f05ca2ee73c3871e94",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01617",
    "objectId": "SYN-PUB-01617"
  }
]

C9 (inference)

R0123 participates in the GENOMFLO project (SYN-PRJ-0254) alongside other listed participants, and co-authored the cited publications with additional authors whose identities are unresolved in the source data; no assumption is made about those authors' affiliation or endorsement of this proposal.

Unresolved participant records mean co-author identities cannot be confirmed.
No affiliation or conflict-of-interest information beyond the profile is available.

Evidence: ev-c7e8de2c805b759e633e2f6328735e9f087252e1884af5370d37272fa10162f5, ev-17eb6d3e27a3376b2f2e3e4d515dba94c575628055a59c5a949a8afe31768b2f, ev-386ebffa1089ce50395b10e140c2bf00d9432f00b34aeb4c55f9d96b8cbee020

[
  {
    "evidenceId": "ev-c7e8de2c805b759e633e2f6328735e9f087252e1884af5370d37272fa10162f5",
    "fieldPointer": "/summary",
    "relation": "source_field",
    "subjectId": "SYN-PRJ-0254",
    "objectId": "SYN-PRJ-0254"
  },
  {
    "evidenceId": "ev-17eb6d3e27a3376b2f2e3e4d515dba94c575628055a59c5a949a8afe31768b2f",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01615",
    "objectId": "SYN-PUB-01615"
  },
  {
    "evidenceId": "ev-386ebffa1089ce50395b10e140c2bf00d9432f00b34aeb4c55f9d96b8cbee020",
    "fieldPointer": "/abstract",
    "relation": "source_field",
    "subjectId": "SYN-PUB-01614",
    "objectId": "SYN-PUB-01614"
  }
]

C10 (fact)

R0123's stated collaboration preferences are joint projects, contract research, consulting and student internships, and prefers partnerships with organisations facing concrete genomic cost or throughput constraints over theoretical or database-curation work.

Preferences are profile statements; actual availability, fees and any institutional IP constraints are not documented.

Evidence: ev-730c8d9381c57c0bd5c7a3f5044e4621d32878f8022ab5fcef6c0158c7fd7baa, ev-51a912e3c4cc6026d4b2e03aa571708f3b6c2b9c32a275d354e47e28c4bfe9db

[
  {
    "evidenceId": "ev-730c8d9381c57c0bd5c7a3f5044e4621d32878f8022ab5fcef6c0158c7fd7baa",
    "fieldPointer": "/open_to",
    "relation": "profile_field",
    "subjectId": "R0123",
    "objectId": "R0123"
  },
  {
    "evidenceId": "ev-51a912e3c4cc6026d4b2e03aa571708f3b6c2b9c32a275d354e47e28c4bfe9db",
    "fieldPointer": "/notes",
    "relation": "profile_field",
    "subjectId": "R0123",
    "objectId": "R0123"
  }
]