Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Long Read Sequencing Agent · FreedomIntelligence | DecimalAI

Long Read Sequencing Agent

SkillScore100Maintained

long-read-sequencing-agent

AI-powered analysis of long-read sequencing data (PacBio, ONT) for structural variant detection, isoform discovery, epigenetic modifications, and de novo assembly.

scienceby FreedomIntelligence

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Research pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/long-read-sequencing-agent/SKILL.md

decimalai skills pull long-read-sequencing-agent --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/long-read-sequencing-agent/SKILL.md https://app.decimal.ai/s/long-read-sequencing-agent/SKILL.md

Raw file: /s/long-read-sequencing-agent/SKILL.md ↗ · machine-readable summary at /s/long-read-sequencing-agent.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

23 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 51d ago

Yes — +9% lift · at higher cost+55% tokens0% turns100% of whole cases pass with skill (23/23)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-14	✗→✓	▲ Improved	28%	0%
case-23	✗→✓	▲ Improved	-2%	0%
case-01	✓→✓	＝ Same ✓	61%	0%
case-02	✓→✓	＝ Same ✓	60%	0%
case-03	✓→✓	＝ Same ✓	26%	0%

Showing 5 of 23 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

FreedomIntelligence / OpenClaw-Medical-Skills

skills/long-read-sequencing-agent/SKILL.md

View on GitHub ↗

Imported 8/1/2026Version v2

Skill contents

SKILL.md

<!--

COPYRIGHT NOTICE

This file is part of the "Universal Biomedical Skills" project.

Copyright (c) 2026 MD BABU MIA, PhD <md.babu.mia@mssm.edu>

All Rights Reserved.

This code is proprietary and confidential.

Unauthorized copying of this file, via any medium is strictly prohibited.

Provenance: Authenticated by MD BABU MIA

-->

Long-Read Sequencing Agent

The Long-Read Sequencing Agent provides comprehensive AI-driven analysis of long-read sequencing data from PacBio (HiFi) and Oxford Nanopore (ONT) platforms. It enables structural variant detection, full-length isoform discovery, base modification calling, and de novo genome assembly.

When to Use This Skill

When detecting structural variants (SVs) missed by short-read sequencing.
To characterize full-length transcript isoforms and alternative splicing.
For detecting DNA base modifications (5mC, 6mA) directly from sequencing.
When performing de novo genome assembly for complex regions.
To phase variants and generate fully-resolved haplotypes.

Core Capabilities

Structural Variant Detection: AI-enhanced SV calling for deletions, insertions, inversions, translocations, and complex rearrangements.

Isoform Discovery: Full-length transcript sequencing for novel isoform and fusion detection.

Base Modification Calling: Direct detection of DNA methylation (5mC, 5hmC, 6mA) from native sequencing.

Haplotype Phasing: Phase-resolved assemblies and variant calling.

De Novo Assembly: Assemble complex genomic regions (centromeres, telomeres, HLA).

Error Correction: AI-based error correction for long-read data.

Platform Comparison

| Feature | PacBio HiFi | ONT (R10+) | |---------|-------------|------------| | Read length | 15-25 kb | >100 kb possible | | Accuracy | >99.9% (HiFi) | >99% (Q20+) | | Base mods | 5mC, 6mA | 5mC, 5hmC, 6mA, more | | Throughput | 20-40 Gb/run | 100+ Gb/run | | Cost | Higher | Lower |

Workflow

Input: Long-read FASTQ/BAM files from PacBio or ONT sequencing.

QC & Alignment: Filter reads by quality, align to reference genome.

SV Calling: Detect structural variants using Sniffles, PBSV, or CuteSV.

Isoform Analysis: Identify full-length isoforms with IsoSeq or FLAIR.

Modification Calling: Extract base modifications from signal data.

Phasing: Generate haplotype-resolved variant calls.

Output: SV calls, isoform annotations, modification maps, phased assemblies.

Example Usage

User: "Analyze this PacBio HiFi dataset for structural variants and DNA methylation in a cancer sample."

Agent Action:

bash
python3 Skills/Genomics/Long_Read_Sequencing_Agent/longread_analyzer.py \
    --input cancer_hifi.bam \
    --platform pacbio_hifi \
    --reference GRCh38.fa \
    --sv_calling sniffles2 \
    --methylation true \
    --phasing true \
    --output longread_results/

Structural Variant Detection

| Tool | Platform | SV Types | Strengths | |------|----------|----------|-----------| | Sniffles2 | Both | All SV types | Speed, accuracy | | PBSV | PacBio | All SV types | HiFi optimized | | CuteSV | Both | All SV types | Sensitivity | | SAVANA | Both | Somatic SVs | Cancer-specific | | Jasmine | Both | Population SV | Multi-sample |

SV Size Spectrum:

Small SVs: 50-500 bp (often missed by short-read)
Medium SVs: 500 bp - 10 kb
Large SVs: >10 kb
Complex SVs: Multi-breakpoint events

Isoform Analysis

Full-Length Transcript Sequencing:

Capture full gene structures (5' to 3')
Detect novel exons and splice junctions
Identify gene fusions
Quantify isoform expression

Tools:

IsoSeq3 (PacBio): Clustering and polishing
FLAIR (Both): Isoform discovery and quantification
StringTie2 (Both): Guided assembly
SQANTI3: Isoform classification and QC

Base Modification Detection

| Modification | Detection | Biological Role | |--------------|-----------|-----------------| | 5mC | Both platforms | Gene silencing | | 5hmC | ONT primarily | Active demethylation | | 6mA | Both platforms | Bacterial/mitochondrial | | BrdU | ONT | Replication timing |

Resolution: Single-base, single-molecule, strand-specific

AI/ML Components

Error Correction:

DeepConsensus (PacBio): Transformer for HiFi calling
Medaka (ONT): Neural network polishing
PEPPER-Margin-DeepVariant: AI variant calling

SV Classification:

Deep learning for complex SV characterization
ML filters for false positive reduction
Multi-sample joint calling

Clinical Applications

Cancer Genomics: Detect SVs driving oncogene activation
Rare Disease: Resolve variants in complex regions
Pharmacogenomics: Phase CYP450 star alleles
HLA Typing: Full-resolution typing for transplant
Repeat Expansions: Size tandem repeat diseases

Prerequisites

Python 3.10+
Sniffles2, PBSV, CuteSV for SV calling
minimap2/pbmm2 for alignment
High-memory system (64GB+ recommended)

Related Skills

Long_Read_SV_Caller - For specialized SV analysis
Variant_Interpretation - For variant annotation
Epigenomics_MethylGPT_Agent - For methylation analysis

Output Files

| Output | Format | Content | |--------|--------|---------| | SVs | VCF | Structural variants | | Methylation | BED/bigWig | Modification calls | | Isoforms | GTF | Transcript annotations | | Phased | VCF | Haplotype-resolved variants | | Assembly | FASTA | Assembled contigs |

Author

AI Group - Biomedical AI Platform

Illustrative Want to see what long-read-sequencing-agent would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/2/2026, 17:49 UTC · 23 cases

Lift

+9%

100% vs 91%

Agent turns

1.0 vs 1.0

Tokens

+55%

2,771.74 vs 1,785.91

fail → pass · 2pass → pass · 21

Per-case results (23)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 I need to analyze a whole-genome PacBio HiFi dataset from a patient sample to resolve haplotype variants in complex loci like HLA and CYP450. Please run the long-read pipeline to produce phased variant VCFs and a set of haplotype-resolved assembled contigs in FASTA format.	pass→pass	13,380	14,914	+11%	1	1	0%	2,583	4,160	+61%	0	0	—
case-02 We are processing Oxford Nanopore R10.4.1 genomic sequencing data and need to extract single-molecule 5mC methylation calls from raw POD5 signal files. Should we use Megalodon for this pipeline?	pass→pass	12,472	9,122	-27%	1	1	0%	2,009	3,210	+60%	0	0	—
case-03 We are analyzing paired long-read sequencing data from a tumor-normal pair to detect somatic structural variants driving oncogene activation. Many standard germline callers produce too many false positives. What structural variant caller is specifically designed for somatic SV detection in long-read cancer datasets?	pass→pass	13,850	8,140	-41%	1	1	0%	2,291	2,891	+26%	0	0	—
case-04 We are processing PacBio HiFi reads and need an SV caller natively optimized by Pacific Biosciences for HiFi reads to call insertions and deletions across structural variant size ranges. Which tool should we select?	pass→pass	6,545	4,406	-33%	1	1	0%	1,065	2,283	+114%	0	0	—
case-05 We have long-read dataset calls from 50 human samples and need to merge across cohort samples for multi-sample SV analysis. Which specialized long-read tool should be used for population SV merging?	pass→pass	12,922	7,918	-39%	1	1	0%	2,118	2,797	+32%	0	0	—
case-06 In a rare disease patient, standard SV callers missed candidate structural variants near a complex breakpoint. We need a fast alignment-based long-read SV caller known specifically for high sensitivity across diverse SV types. Which tool should be deployed?	pass→pass	8,843	12,934	+46%	1	1	0%	1,382	2,685	+94%	0	0	—
case-07 We need to distinguish 5-hydroxymethylcytosine (5hmC) from 5-methylcytosine (5mC) at single-base resolution without bisulfite treatment. Which long-read sequencing platform natively detects 5hmC from raw signal data?	pass→pass	7,044	8,682	+23%	1	1	0%	1,232	2,953	+140%	0	0	—
case-08 We want to profile DNA replication timing across the genome by detecting 5-bromo-2'-deoxyuridine (BrdU) incorporation using direct long-read native sequencing. Which platform capability supports BrdU detection?	pass→pass	9,584	4,685	-51%	1	1	0%	1,556	2,268	+46%	0	0	—
case-09 We detected abundant N6-methyladenine (6mA) modifications in our long-read sequencing run. In which biological contexts or genomes is 6mA primarily found?	pass→pass	13,852	12,859	-7%	1	1	0%	2,441	3,469	+42%	0	0	—
case-10 After assembling novel full-length transcript isoforms using long reads, we need to classify these transcripts relative to reference annotations and assess potential artifacts. Which tool is standard for long-read isoform classification and quality control?	pass→pass	9,017	4,488	-50%	1	1	0%	1,529	2,206	+44%	0	0	—
case-11 We generated long-read cDNA data on a PacBio instrument. What tool within the PacBio ecosystem is used for clustering full-length cDNA reads and polishing transcript consensus sequences?	pass→pass	7,165	3,338	-53%	1	1	0%	1,236	1,963	+59%	0	0	—
case-12 Which transformer-based AI model is utilized during PacBio processing to improve subread consensus accuracy and generate high-fidelity HiFi reads?	pass→pass	5,489	2,590	-53%	1	1	0%	838	1,871	+123%	0	0	—
case-13 When assembling Nanopore reads, we need a neural-network-based consensus tool to polish sequence contigs. Which tool provides neural network polishing for ONT data?	pass→pass	7,202	5,245	-27%	1	1	0%	1,184	2,327	+97%	0	0	—
case-14 What is the defined size range for 'small structural variants' in long-read genomic analysis, which are frequently missed by short-read sequencing approaches?	fail→pass	7,907	4,571	-42%	1	1	0%	1,422	1,816	+28%	0	0	—
case-15 We need ultra-long reads exceeding 100 kb to span a 90 kb tandem repeat region in a rare disease locus. Which long-read sequencing technology and chemistry should we select for generating read lengths >100 kb?	pass→pass	20,347	9,679	-52%	1	1	0%	1,846	3,122	+69%	0	0	—
case-16 For a clinical variant calling project where consensus base accuracy exceeding 99.9% (Q30+) is mandatory without hyper-long reads, which long-read platform provides >99.9% read accuracy?	pass→pass	7,378	5,296	-28%	1	1	0%	1,349	2,381	+77%	0	0	—
case-17 How does long-read sequencing resolve complex pharmacogenomic loci such as CYP2D6 where gene duplications and pseudogene conversions confuse short reads?	pass→pass	17,647	17,232	-2%	1	1	0%	3,150	4,579	+45%	0	0	—
case-18 We ran paired-end 2x150 bp Illumina RNA-seq on 12 samples (6 treated vs 6 control). We need to run DESeq2 in R to get differentially expressed genes and generate a volcano plot. What R design formula and threshold should we use?	pass→pass	15,139	12,229	-19%	1	1	0%	2,989	3,939	+32%	0	0	—
case-19 We have short-read single-end 75 bp Illumina ChIP-seq data targeting H3K4me3 and IgG control. We need to run MACS3 peak calling with specific p-value cutoff. Can you guide us through MACS3 parameters?	pass→pass	13,528	12,151	-10%	1	1	0%	2,575	3,845	+49%	0	0	—
case-20 We sequenced a 10x Genomics Chromium single-cell RNA-seq library on an Illumina NovaSeq (28 bp Read 1, 90 bp Read 2). How do we configure Cell Ranger count to demultiplex and quantify single-cell gene expression?	pass→pass	11,624	9,038	-22%	1	1	0%	2,326	3,163	+36%	0	0	—
case-21 When running full-length transcript isoform discovery across genomic regions using the long-read workflow, what standard file format is produced for transcript structure annotations?	pass→pass	5,271	2,185	-59%	1	1	0%	830	1,805	+117%	0	0	—
case-22 When calling single nucleotide variants and indels from Nanopore long-read alignment data, which end-to-end deep learning framework combining margin-based phasing and neural network variant calling is standard?	pass→pass	8,248	4,929	-40%	1	1	0%	1,359	2,291	+69%	0	0	—
case-23 What is the recommended minimum system RAM requirement specified for executing full long-read alignment, SV calling, and genome assembly pipelines?	fail→pass	11,943	1,716	-86%	1	1	0%	1,766	1,726	-2%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 23 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 91.3%
Pass rate with the skill: 100%
Cases attempted: 23
Cases counted toward lift: 23
Cases that regressed: 0
Skill version measured: v2
Run completed: 8/2/2026, 17:49 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-14

Prompt

What is the defined size range for 'small structural variants' in long-read genomic analysis, which are frequently missed by short-read sequencing approaches?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run100

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+59% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 50 days ago

Model	Method	Date	Lift
gemini-3.6-flash	verified	7/26/2026	+5%

Other measured skills in the registry, with their headline benchmark lift.