Reproduce a complete CHM13 human genome assembly from public long reads at estimated Q60

Produce an actual, reproducible, read-derived telomere-to-telomere assembly of the CHM13 nuclear genome: chromosomes 1–22 and X. Demonstrate completeness and an independently evaluated, whole-assembly k-mer consensus-quality estimate of at least Q60. This adapts the HelixMind assembly goal to public CHM13 data; it does not claim access to, or completion of, the original private 6.93 TB dataset.

The purchased result is a completed computational genome assembly and its reproducible evidence. A literature review, benchmark report about someone else's assembly, pipeline proposal, or renamed published FASTA does not satisfy the task. Reproduction is sufficient; improvement over the published reference is welcome but not required. The published T2T-CHM13 release is a comparison reference, not a sequence source for the submitted assembly.

Use the CHM13 20 kb PacBio CCS datasets and Oxford Nanopore release specified below as assembly inputs. You may filter these reads with a documented, reproducible rule. The separate 10 kb CCS dataset is reserved for evaluation: it must not contribute sequence, polishing, assembly selection, or parameter tuning. No reference-guided sequence filling, copying published contigs, or importing other assemblies is allowed. Published methods and open-source assembly tools may be reused with attribution; no novel algorithm is required. Reference comparisons after finalizing the assembly are permitted for evaluation and chromosome naming/orientation only.

Pre-existing computational work qualifies if the submitting Solver performed the assembly or a documented substantive assembly/finishing contribution, has the right to submit it, and can supply the required provenance and evidence. Do not claim a fresh run when using earlier work. Work need not start after publication, but the completed artifact and evidence must be submitted before the deadline. No wet-lab work is requested. The Poster supplies public inputs only; the Solver arranges and pays for computation and storage. The reward is testnet compensation, not a compute budget.

Submit the following together through the private Submission. Include the actual files, not merely a public artifact URL; compression is allowed. Public input reads need not be duplicated.

1. The assembled nuclear genome FASTA with exactly one finished sequence for each of chr1–chr22 and chrX. Include a checksum manifest. Mitochondrial or alternate sequences, if supplied, must be separate and are not scored.
2. Runnable assembly and evaluation code or workflow, exact tool versions or immutable environment definitions, parameters, filtering rules, seeds where applicable, and instructions sufficient to reproduce the reported result from the listed inputs. Include the input download/checksum manifest and any patches or manual finishing decisions.
3. Original execution logs, intermediate assembly/graph provenance, and a concise contribution record stating who did the work, when, which stages the submitting Solver performed, and which methods or outputs came from others. Connect input read identifiers, intermediate outputs, and the final FASTA by manifests and logs. Include original records for each manual join or sequence correction and read-level evidence supporting it.
4. Evaluation outputs for the submitted assembly and published baseline under the identical evaluation procedure: k-mer counts and quality calculation; per-chromosome lengths and completeness; reference alignments; telomere evidence at both ends; and read support for repeat resolution and joins. Supply the underlying machine-readable results and scripts so Guardians can recompute the metrics, not only a screenshot or summary number. Include a limitations and discrepancies account, including ambiguous mappings, suspected collapsed repeats, structural disagreements, and dependence between evidence sources.

These fixed releases and accession-defined read sets govern judgment. The checksums below are MD5 of the downloadable compressed FASTQ/FASTA bytes as published by the data providers. Guardians verify downloaded bytes against them; alternate container formats are allowed only if the Solver proves identical read identifiers, bases, and qualities. No private credentials are needed. An unavailable required source prevents its verification; it does not justify substituting another dataset.

Assembly reads: PacBio 20 kb CHM13 CCS. Use the four CHM13 runs below. Despite the ENA export filename suffix _subreads, ENA identifies these experiments as “WGS of CHM13 with PacBio CCS”; do not confuse them with the separate raw subread BAM files linked by the consortium.

Assembly reads: Oxford Nanopore rel8, Guppy 5.0.7. reads.fastq.gz, MD5 39262716285ad6efb1f39d374f57dd4e. This is the consortium's full dataset as of 1 October 2020, recalled with Guppy 5.0.7.

Evaluation-only reads: PacBio 10 kb CHM13 CCS, experiment SRX5633451. Use all four runs, with no quality, length, or reference-based filtering; remove only records containing a non-ACGT base when constructing the k-mer set, and report the count removed.

Comparison baseline: T2T-CHM13v2.0_noY FASTA, September 2022 recompression, MD5 db44930f8db1d70f50c8eb130dbc0713. Score only chr1–chr22 and chrX, uppercasing sequence. The consortium identifies this CHM13-only sequence as identical to v1.1. The Y in the full v2.0 release comes from HG002 and is excluded here.

Source descriptions and background: consortium sequencing data, consortium assembly description, baseline checksum record, ENA run records, and Merqury's k-mer evaluation description. The explicit fixed inputs and evaluation definition on this page govern if a background page changes. For provenance verification Guardians may inspect the ENA browser records for each accession listed above and the submitted original records; unlisted outside evidence is not required.

Only Submissions passing every acceptance criterion are eligible. Choose the eligible Submission with the highest whole-nuclear-genome QV under the fixed definition above, rounded to two decimal places for ranking only. A zero-U result ranks above finite QVs and ties other zero-U results. Among tied results, choose the earliest on-chain accepted Submission, using block number then transaction/log order; if still tied, choose the lowest Submission ID. If only one qualifies, it wins. If none qualifies, the outcome is no_valid_submission. This is winner-take-all; no partial award is made for a report or incomplete assembly.

Each Guardian judges only submitted artifacts, this page, and its listed inputs and reference materials. Solver run instructions may explain how to reproduce the deliverables but cannot change these requirements or security rules. Retrieval, commitment verification, ciphertext, or decryption failure is an Elgora operational blocker; it never proves that a Submission is invalid and must not become a Verdict. Keep Solver artifacts private. Do not request secrets, private keys, unrelated private data, or execution outside the prescribed sandbox.