Independent Scientific Consultancy

Krzysztof
Gizynski

Molecular Diagnostics · Microbial Genomics · Computational Biology

I help biotechnology and diagnostics teams select and evaluate diagnostic targets, support molecular assay development, and build reproducible genomics workflows — combining experimental experience with computational analysis.

Connecting Molecular Diagnostics and Computational Biology

I am a molecular biologist with a microbiology background and industry experience in molecular diagnostics, assay development and biotechnology.

Alongside that work, I have built independent computational biology projects: reproducible bioinformatics workflows for microbial genomics, metagenomics and diagnostic applications, extended more recently to human clinical genomics — variant calling, ACMG/AMP classification, pharmacogenomics and somatic tumour–normal analysis.

Having worked on both sides, I can connect what an assay has to do in practice with what the sequence data can and cannot show. The case studies below are self-directed projects, with code on GitHub, that show how I work.

833 Bacterial genomes processed across projects
4 Pathogens profiled
DSL2 Nextflow DSL2 — used across the Nextflow-based projects
AMR Resistance surveillance focus
1.0/1.0 GIAB HG002 recall/precision (clinical pipeline)
8/8 Somatic PASS calls confirmed (COLO829 tumour–normal)
38 Real GeT-RM (CDC) reference samples validated (PGx, 5 genes)
3 Inheritance patterns handled (AR · X-linked · AD) in the ACMG/AMP classifier

How I Can Help

I work with biotechnology, diagnostics and life-science teams on problems where molecular-biology insight and reproducible computational analysis both matter.

Molecular Diagnostics & Microbial Genomics

Computational support for diagnostic assay design and pathogen genomics.

  • Diagnostic target selection and evaluation
  • Inclusivity and exclusivity analysis
  • Assay-development support, including in-silico primer/probe design and QC
  • AMR profiling and pathogen typing
  • Metagenomic profiling, AMR surveillance and analytical method benchmarking
  • Phylogenomics and genome-collection curation

Clinical Genomics & Pipeline Development

Reproducible, benchmarked analysis workflows for human genomic data.

  • Germline and somatic variant-calling workflows
  • Benchmarking against reference truth sets (GIAB, GeT-RM)
  • ACMG/AMP interpretation workflows
  • Pharmacogenomics workflows
  • Nextflow pipelines and containerised, reproducible analysis

My work is computational and advisory. I do not provide clinical interpretation of patient results or regulated diagnostic reporting. Engagements are scoped project by project — get in touch to discuss yours.

Case Studies

Independent projects that show the methods behind the services above. Each links to its code on GitHub.

01Metagenomics & Analytical BenchmarkingBenchmarking taxonomic and resistance-gene profiling, reanalysing published microbiome data, and testing whether a published analysis can be reproduced.

Nanopore Shotgun Metagenomics · Benchmark

Pathogen & AMR Profiling Benchmark

GitHub

Benchmark of taxonomic profiling and antimicrobial-resistance gene detection on public Nanopore shotgun data from a defined gut mock community (Zymo D6331, PRJNA1181840). Experiments were registered before they were run, tuned on a development read pool, frozen with a git tag, and scored once on a separate evaluation pool; every arm and deviation is reported, including the target that was missed. Research use only.

Kraken2 and Sylph were assessed against the community’s theoretical composition, including sequencing-depth, confidence-threshold and abundance comparisons. Resistance-gene reads were then tested for correct assignment to their bacterial carriers. Species detection was strong, but species attribution missed its registered target. The sequenced kit lot could not be confirmed, limiting interpretation against the theoretical reference composition.

0.966 species F1 on the evaluation pool (14 TP · 1 FP · 0 FN)
14/17 species evaluable at a 20-read minimum
0.953–0.966 F1 across 0.1 / 0.25 / 0.5 Gb depth
3 resistance genes detected: cepA · tet(Q) · tet(W)
0.593 species attribution vs 0.85 target — not met
Nextflow DSL2 Kraken2 Sylph minimap2 ResFinder AMRFinderPlus Docker Python
View Workflow Diagram
Nanopore Metagenomics Benchmark Workflow A public Nanopore run of a gut mock community is split into development and evaluation read pools, experiments are registered and the pipeline frozen, then taxonomic profiling and resistance-gene detection are scored against the theoretical composition, and every result is reported including a missed target. Nanopore Metagenomics Benchmark Registered experiments · dev/eval split · scored against the theoretical composition Public Nanopore run SRR31222153 Zymo D6331 gut mock community (PRJNA1181840) DEV / EVAL READ POOLS tune on dev, score eval once REGISTER, THEN FREEZE plan registered · git tag before eval TAXONOMIC PROFILING Kraken2 PlusPF-8 · Sylph RESISTANCE GENES minimap2 → ResFinder · AMRFinderPlus catalogue Species F1 0.966 14 TP · 1 FP (human) · 0 FN · 14 of 17 evaluable cepA · tet(Q) · tet(W) detected 22 · 51 · 16 supporting reads Depth series · confidence · abundance F1 0.953–0.966 · Bray-Curtis 0.137–0.168 Species attribution of AMR reads 0.593 vs 0.85 target — not met All arms and deviations reported Nextflow DSL2 core path · eval refused without --allow_eval · images pinned by digest Analysis step Key output Target not met Reporting
Wastewater Shotgun Metagenomics · Preregistered Reproduction

Resistance Genes by Metagenomics vs qPCR

GitHub

Reproduced the metagenomic analysis of a published wastewater study (Taylor et al., 2025; PRJNA904380) using public sequencing reads, reported methods and documented assumptions. Nine Day 1 liquid samples were used for development; the criteria and scoring script were then frozen before 18 held-out liquid samples were evaluated once. The study examined read counts for five resistance genes and compared detection calls with the authors’ published qPCR findings. Research use only.

All registered reproduction criteria were met, but ermB counts differed substantially in three effluent samples and the published ARG/16S scale could not be reproduced. Exploratory analyses supported competition between closely related references as a factor in KMA read assignment, although the exact cause of the difference from the authors remains unresolved. Gene-level agreement was stronger than subtype-level agreement; the qPCR comparison was descriptive.

17/18 samples with summed ARG read counts within ±7% of the published sum (registered: 16/18)
68/70 gene–sample pairs within ±10% of the published count
0 additional gene detections (registered limit: 2)
165/165 sample–gene presence calls match the paper’s Table S5
3/9 effluent samples with low ermB — reported, cause unresolved
Preregistration KMA bbduk ResFinder SILVA conda Bash Python
View Workflow Diagram
Wastewater Resistance-Gene Metagenomics Reproduction Workflow Public shotgun reads from a wastewater study are split into development and evaluation samples, the criteria and scoring script are registered and frozen, then reads are trimmed and aligned to a resistance-gene database and a 16S database and scored against the published counts. The criteria were met, while ermB in three effluent samples and the 16S scale were not reproduced. Wastewater ARG Metagenomics: Preregistered Reproduction Public reads · develop on Day 1 · freeze · score 18 held-out samples once Public shotgun reads PRJNA904380 33 wastewater samples from one plant (Taylor et al. 2025) DEVELOPMENT / EVALUATION SPLIT Day 1: 9 develop · Days 2–3: 18 evaluate · 6 sediment descriptive REGISTER, THEN FREEZE criteria and scoring script registered · git tag before evaluation TRIM + RESISTANCE GENES bbduk → KMA vs ResFinder 2.1.1 · gene and subtype counts 16S DENOMINATOR KMA vs SILVA 138.1 NR99 · ARG per 16S read Registered criteria P1–P4 met P1 17/18 · P2 18/18 · P3 68/70 · P4 17/18 · 0 extra genes ARG per 16S not reproduced about 100× smaller than the paper’s scale · unit unverified ermB low in 3 of 9 effluent samples reference competition supported · cause unresolved Detection vs qPCR (exploratory) Table S5: 165/165 · 26/36 qPCR positives found in pond + sediment All results and post-freeze changes reported gene-level agreement stronger than subtype-level · limits stated · MIT-licensed code Analysis step Criteria met Not reproduced Reporting
16S Amplicon Microbiome · Prespecified Reanalysis

Diet and gut microbiome recovery after antibiotics — a 16S reanalysis

GitHub

Reanalysed published 16S rRNA data from humanised mice to compare gut microbiome recovery after ciprofloxacin under standard and low-MAC diets. Built a QIIME 2 workflow using DADA2 and SILVA, with a prespecified analysis plan, cage-level summaries and sensitivity analyses. Reproduced the primary result in nf-core/ampliseq with matched settings.

Low-MAC cages were closer to their own pre-antibiotic composition during days 8–12 by Bray–Curtis distance. Baseline samples for the two diets were sequenced in different runs, which limits attributing this pattern to diet.

−0.145 low-MAC − standard Bray–Curtis distance from day 0 (cage-equal)
2/2 low-MAC cages below all 3 standard-diet cages; mice overlap
−0.1453 vs −0.1454: nf-core/ampliseq vs QIIME 2 (prespecified tolerance 0.02)
453 samples demultiplexed · 451 passed initial QC · 17 mice across 5 cages in the primary analysis

Original data: Ng KM, Aranda-Díaz A, Tropini C, et al. Recovery of the gut microbiota after antibiotics depends on host diet, community context, and environmental reservoirs. Cell Host & Microbe. 2019;26(5):650–665.e4. doi:10.1016/j.chom.2019.10.011
Sequence data: Stanford Digital Repository, purl.stanford.edu/rm696bw4911, licensed CC BY-NC 3.0.
Analysis and figures: Krzysztof Gizynski, independently reanalysing the published dataset of Ng et al. (2019).

QIIME 2 DADA2 SILVA q2-boots nf-core/ampliseq Nextflow Docker Python
View Workflow Diagram
16S Microbiome Reanalysis Workflow Public 16S reads from a humanised-mouse study are demultiplexed once per run, the analysis plan is frozen, and the same per-sample reads are processed in QIIME 2 and, as a reproducibility check, in nf-core/ampliseq with matched settings. The primary result was reproduced; it depends on the distance metric, and a sequencing-run confound between diets is documented. 16S Reanalysis: Diet and Recovery after Antibiotics Public reads · frozen plan · QIIME 2 · matched second workflow · cage-level reporting Public 16S V4 reads (Ng et al. 2019) humanised mice · 3 MiSeq runs · 453 samples · rm696bw4911 DEMULTIPLEX ONCE PER RUN EMP paired-end, Golay correction · checksummed per-sample FASTQs FREEZE THE ANALYSIS PLAN endpoint · 17 eligible mice · cage weighting · sensitivity set · git tag QIIME 2 WORKFLOW DADA2 per run · SILVA · 100 rarefactions at 6,400 reads nf-core/ampliseq 3.1.0 DADA2 settings matched from source · rule tagged before run Frozen primary result low-MAC − standard = −0.145 · 2 cages below all 3 Result reproduced −0.1453 vs −0.1454 · same cage pattern · tolerance 0.02 Depends on the distance metric stable across Bray–Curtis versions · UniFrac, Aitchison overlap Design confound baselines in different runs by diet · concurrency unverified Diet effect not claimed descriptive cage-level comparison · baseline run confounded with diet Analysis step Result Limit Reporting

02Molecular Diagnostics & Microbial GenomicsAssay design, resistance profiling, typing and phylogenomics of bacterial pathogens.

Neisseria gonorrhoeae

AMR Diagnostic Pipeline

GitHub

End-to-end Nextflow DSL2 pipeline characterising antimicrobial resistance across 283 clinical strains. Integrates AMRFinderPlus resistance profiling with a custom alignment-based mutation caller (penA, gyrA, parC, mtrR, porB) and promoter-variant detection, core-genome phylogeny via Parsnp, and an interactive Dash dashboard linking resistance phenotypes to phylogenetic context for exploratory analysis and antimicrobial-resistance surveillance.

283 strains
Reference: FA1090
60% ciprofloxacin resistance (GyrA QRDR)
51% reduced cephalosporin susceptibility (mosaic penA)
Mean 10.6 AMR elements per strain
Dual-track: AMRFinderPlus + custom mutation caller
mtrR · porB promoter-variant analysis
Core-genome phylogeny
Interactive dashboard
Nextflow DSL2 AMRFinderPlus Parsnp BLAST+MAFFT Dash Docker Python
View Workflow Diagram
N. gonorrhoeae Diagnostic Pipeline Workflow Two-stream pipeline diagram: AMR and promoter analysis, and core genome phylogeny, converging into a Dash dashboard. N. gonorrhoeae Diagnostic Pipeline 283 clinical strains · FA1090 reference · Nextflow DSL2 · Docker 283 N. gonorrhoeae Genomes NCBI RefSeq · complete assemblies Prokka Annotation GFF · protein FASTA ① AMR Profiling ② Core Genome Phylogeny AMRFinderPlus Resistance genes · point mutations QRDR Mutation Calling penA · gyrA · parC · mtrR · porB BLAST+MAFFT alignment Promoter Variant Analysis mtrR · porB · IR detection Profile Aggregation Per-strain · cross-strain summary Resistance Profiles Per-strain TSV · 54 elements · 8 drug classes Parsnp Core-genome SNP alignment SNP Distance Matrix Pairwise distances · clustering ML Phylogeny FastTree · SNP-based tree Clade Annotation 4 major clades identified AMR Burden Overlay Clonality vs convergence Interactive Dash Dashboard 5 tabs · AMR overview · Mutations · Phylogeny · Promoters Analysis step Key output Dashboard
Streptococcus pyogenes

AMR & Typing Pipeline

GitHub

Scalable pipeline processing 375 publicly available S. pyogenes genomes with integrated resistance profiling, MLST, and emm typing. Identified 22.9% AMR prevalence across the cohort, with ST28 as the dominant sequence type (22.4%) and emm5 the most prevalent emm type (47.5%). Designed for rapid integration of newly sequenced clinical isolates into ongoing surveillance.

375 genomes
22.9% AMR prevalence
ST28 dominant (22.4%)
emm5 dominant (47.5%)
Nextflow DSL2 AMRFinderPlus MLST emmtyper Docker Python
View Workflow Diagram
S. pyogenes AMR and Typing Pipeline Workflow Three-stream pipeline: AMR profiling, MLST typing, and emm typing of 375 S. pyogenes genomes, converging into an integrated report. S. pyogenes AMR & Typing Pipeline 375 clinical strains · pubMLST spyogenes scheme · Nextflow DSL2 · Docker 375 S. pyogenes Genomes NCBI RefSeq · complete assemblies ① AMR Profiling ② MLST Typing ③ emm Typing AMRFinderPlus Resistance genes · Streptococcus_pyogenes Resistance Classification mef · erm macrolide · tet genes POINT Mutation Screening --plus flag · no beta-lactam resistance detected Profile Aggregation Per-strain · cross-strain summary AMR Prevalence: 22.9% Macrolide · tetracycline dominant MLST 7-locus gki · gtr · murI · mutS · recP · xpt · yqiL Allele Profiling PubMLST spyogenes database ST Assignment Sequence type per genome ST Frequency Distribution 133 unique STs · ranked by prevalence ST28 Dominant: 22.4% 133 distinct STs identified emmtyper In silico emm gene typing emm Type Assignment CDC emm database Prevalence Ranking Type frequency analysis AMR × emm Cross-tabulation AMR carriage rate per emm type emm5 Dominant: 47.5% 11 emm types identified Integrated AMR + Typing Report Per-strain TSV · cohort summary · 375 genomes · 47 min runtime Analysis step Key output
Chlamydia trachomatis · Assay Design

In-silico validated Workflow for Diagnostic Target Selection

Multiplex qPCR/TaqMan In-silico Design

Reproducible target-selection workflow for a multiplex qPCR/TaqMan assay for C. trachomatis. Genome QC and tiering are performed before target discovery, ensuring that consensus sequences and inclusivity assessments are built only from high-confidence assemblies. A provenance audit traced 149 GenBank-only genomes to a single study of engineered laboratory strains; these were excluded from the diagnostic reference set, which was rebuilt from 125 complete RefSeq assemblies that were independently verified and tiered.

Four chromosomal loci (mutL, aroB, PSD2, and group_306) were independently evaluated as candidate targets alongside the multicopy diagnostic plasmid, which serves as the higher-sensitivity target. The proposed duplex pairs one chromosomal locus with the plasmid target, combining genomic redundancy with sensitivity.

The workflow integrates pan-genome analysis, sequence-conservation and inclusivity screening, exclusivity BLAST, quantitative Tm-gap assessment, and three-way structural QC of primer–probe sets, producing a computationally screened candidate panel for downstream experimental validation.

Chromosomal targets — mutL · aroB · PSD2 · group_306
125 genomes (97 primary / 10 review / 18 excluded)
149 GenBank-only engineered-strain genomes rejected
874 core genes (Panaroo pan-genome)
692/874 exclusivity-clean
4 targets · 7 oligo sets
Plasmid target — Primary + Reserve windows
34 plasmid-bearing genomes
Circular-alignment artifact caught & fixed (7,606 bp)
1 plasmid target · 2 candidate windows · 6 oligos
Tm gap 6.08–8.07°C across all 9 sets (≥5°C floor)
Panaroo MAFFT fastANI primer3-py NCBI BLAST Three-way dimer QC Python
View Workflow Diagram
C. trachomatis Multiplex Assay Design Workflow Two parallel streams from the same organism — chromosomal targets (125 genomes) and the diagnostic plasmid (34 genomes) — each processed through pan-genome/consensus analysis, exclusivity screening, Primer3 design, and three-way structural QC, converging into one locked multiplex panel. C. trachomatis Multiplex Assay Design Inclusivity · Exclusivity · Tm-Gap Rule · Three-Way Dimer QC · Primer3 Two genomic compartments, one QC standard ① Chromosomal Targets ② Plasmid Target 125 C. trachomatis Genomes RefSeq only · 149 engineered GenBank genomes rejected Genome QC & Tiering 97 primary · 10 quality-review · 18 excluded Panaroo Pan-genome 904 clusters → 874 core (≥99%) Exclusivity BLAST 692/874 core genes exclusivity-clean Primer3 Design Tm-gap ≥5°C rule · real-computed Tm Three-Way Dimer + Structure QC primer3-py hairpin/homo/heterodimer mutL · aroB · PSD2 · group_306 4 targets · 7 oligo sets 34 Plasmid-Bearing Genomes Circular contig, 7,415–7,510 bp Rotation-Corrected Alignment Fixed 2×-length MAFFT artifact Consensus Build 7,606 bp corrected consensus Primer3 Design Same Tm-gap ≥5°C standard Three-Way Dimer + Structure QC Same primer3-py standard BLAST Specificity Human genome + core_nt · pair-level rule Primary + Reserve Windows One plasmid · 2 candidate windows · 6 oligos Validated Diagnostic Candidate Targets 5 targets validated (4 chromosomal + 1 plasmid) · 9 oligo sets · 27 oligos Analysis step Key output
Klebsiella pneumoniae

Phylogenomics Pipeline

GitHub

Comprehensive phylogenomics pipeline spanning the full analysis chain for 50 K. pneumoniae genomes. Assembly QC (QUAST), AMR and resistance gene profiling (AMRFinderPlus), virulence and typing (Kleborate), SNP-based variant calling (Snippy), recombination masking (Gubbins), and maximum-likelihood phylogeny (IQ-TREE). Results delivered via a three-tab Dash dashboard. ST11 identified as the dominant lineage, consistent with global carbapenem-resistant K. pneumoniae trends.

50 genomes
ST11 dominant
3-tab dashboard
Full QUAST → IQ-TREE chain
Nextflow DSL2 QUAST AMRFinderPlus Kleborate Snippy Gubbins IQ-TREE Dash Docker
View Workflow Diagram
K. pneumoniae Phylogenomics Pipeline Workflow Three-stream pipeline: AMR profiling, molecular typing, and core genome phylogeny of 50 K. pneumoniae genomes, converging into a three-tab Dash dashboard. K. pneumoniae Phylogenomics Pipeline 50 clinical genomes · Pasteur MLST scheme · Nextflow DSL2 · Docker 50 K. pneumoniae Genomes NCBI RefSeq · complete assemblies QUAST Quality Control N50 · contig count · assembly size ① AMR Profiling ② Molecular Typing ③ Core Genome Phylogeny AMRFinderPlus Resistance genes · Klebsiella_pneumoniae β-lactamase Detection ESBL · KPC · OXA · NDM classes Resistance Gene Matrix Per-strain binary presence/absence Drug Class Summary Cross-strain aggregation Resistance Profiles ESBL · KPC · carbapenem prevalence Kleborate MLST · virulence loci · capsule ST Assignment Pasteur scheme · 7 loci Virulence Scoring ybt · clb · iuc · iro · rmpA loci Clonal Group Analysis CG258 · high-risk lineages ST11 Dominant ST258 · ST307 · ST147 · ST16 Snippy SNP calling vs. reference genome Gubbins Recombination masking IQ-TREE Maximum likelihood · GTR+G model snp-dists Pairwise SNP distance matrix Phylogenetic Tree Clonal clusters · AMR burden overlay Interactive Dash Dashboard 3 tabs · Phylogeny+ST · AMR Heatmap · Virulence Heatmap Analysis step Key output Dashboard

03Human & Clinical GenomicsVariant calling, ACMG/AMP classification, pharmacogenomics and somatic analysis.

CAPN3 / DMD

Variant Calling Pipeline

GitHub

Reads-to-VCF pipeline scoped to the CAPN3 and DMD loci, independently benchmarked against Genome in a Bottle's HG002 reference sample. Raw reads are QC'd (FastQC/MultiQC) and adapter/quality-trimmed (fastp) before alignment. Two variant callers — GATK HaplotypeCaller and DeepVariant — are run independently and cross-checked against each other, then the concordant call set is annotated with Ensembl VEP transcript consequence and gnomAD population frequency, producing evidence ready for the downstream ACMG classifier.

1.0/1.0 SNP + INDEL recall/precision (GIAB HG002)
11-module Nextflow DSL2 pipeline
Two callers cross-checked: GATK + DeepVariant
2,267 concordant calls annotated (VEP + gnomAD v4.1)
fastp-trimmed reads, FastQC/MultiQC QC'd pre- and post-trim
Remote HTTP-range extraction — no full-genome download
CAPN3 (chr15) + DMD (chrX) scoped
Nextflow DSL2 FastQC fastp BWA-MEM2 GATK DeepVariant VEP gnomAD hap.py Docker
View Workflow Diagram
CAPN3/DMD Variant Calling Pipeline Workflow Reads-to-VCF pipeline: raw-read QC and adapter/quality trimming, then alignment and dual-caller variant calling converging into cross-check, then forking into GIAB benchmark and VEP/gnomAD annotation, feeding the ACMG classifier. CAPN3/DMD Variant Calling Pipeline CAPN3 (chr15) + DMD (chrX) · GIAB HG002 benchmark · Nextflow DSL2 · Docker HG002 GRCh38 BAM (remote) GIAB · HTTP range-extracted, no full download EXTRACT_REGION region FASTQ (R1/R2) FASTQ_QC (raw) FastQC + MultiQC · pre-trim TRIM_READS fastp · adapter/quality trim FASTQ_QC (trimmed) FastQC + MultiQC · post-trim BWA_ALIGN BWA-MEM2 → aligned BAM SORT_MARKDUP sort + mark duplicates (GATK) ① GATK HaplotypeCaller ② DeepVariant GATK_CALL HaplotypeCaller · 2,314 variants DEEPVARIANT_CALL WGS model · 2,906 variants CROSS_CHECK_VCFS bcftools isec · 2,267 concordant calls ③ Benchmark (hap.py) ④ Annotate (VEP + gnomAD) HAPPY_BENCHMARK vs. GIAB HG002 truth (NISTv4.2.1) ANNOTATE_CALLS VEP transcript + gnomAD v4.1 AF 1.0/1.0 recall/precision SNP + INDEL, both callers vs. GIAB 2,267 annotated calls concordant set · VEP + gnomAD → CAPN3-DMD-variant-classifier annotated VCF → VariantEvidenceBundle Analysis step Key output Feeds classifier
CAPN3 / DMD / BRCA1

ACMG/AMP Variant Classifier

GitHub

A scoped ACMG/AMP (Richards et al. 2015) variant classification engine, now spanning three genes and two inheritance mechanisms: CAPN3 (limb-girdle muscular dystrophy, autosomal recessive) and DMD (Duchenne/Becker muscular dystrophy, X-linked) form the original scope, extended to BRCA1 (hereditary breast/ovarian cancer syndrome, autosomal dominant with incomplete penetrance) as a deliberate generalization test against the real ClinGen ENIGMA BRCA1/BRCA2 VCEP v1.2 specification. Applies twelve ACMG criteria across population frequency, protein effect, computational evidence, functional-assay data, and structural variation — including a dedicated CNV deletion/duplication scoring module for DMD's largely exon-deletion mutational spectrum — combined by two independent systems (classic Table 5 and Tavtigian et al. 2020's Bayesian points) cross-checked against each other, plus a case-level layer reasoning about whether a patient's variant(s) actually explain their disease: biallelic X-linked (XX) inheritance scenarios for DMD, and a new "risk-conferring" case status reflecting BRCA1's dominant, incompletely penetrant mechanism rather than reusing the deterministic status the other two genes use.

35 point-mutation + 5 CNV + 14 case-level fixtures
259/259 automated tests passing
Three genes, two inheritance mechanisms (CAPN3 AR · DMD X-linked · BRCA1 AD)
Twelve ACMG/AMP criteria evaluators
Two combining systems (Table 5 + Bayesian, cross-checked)
CNV deletion + duplication scoring module (DMD)
Case-level reasoning (AR + X-linked incl. biallelic XX + BRCA1 risk-conferring)
Fed directly by the variant-calling pipeline's annotated output
Python dataclasses ACMG/AMP CNV PyYAML stdlib-only
View Workflow Diagram
CAPN3-DMD-variant-classifier Workflow ACMG/AMP evidence bundle evaluated against twelve criteria, with a separate CNV deletion/duplication module branching off in parallel, combined by two independent systems, cross-checked, then reasoned about at the case level. CAPN3-DMD-variant-classifier 35 point + 5 CNV + 14 case-level fixtures · 259/259 tests · twelve ACMG/AMP criteria · 3 genes VariantEvidenceBundle curated evidence, or fed by the calling pipeline Twelve ACMG/AMP Criteria Evaluators population frequency · protein effect · computational scores ① Classic (Table 5) ② Bayesian Points ③ CNV (DMD structural variants) TABLE5_COMBINE Richards et al. 2015 BAYESIAN_COMBINE Tavtigian et al. 2020 CNV_SCORING Riggs et al. 2020 Cross-checked classification Pathogenic → Benign · 2 systems agree CNV classification VUS → Pathogenic · Section 2 (dosage) Case-Level Reasoning AR (CAPN3) · X-linked (DMD) · AD risk-conferring (BRCA1) Clinical Interpretation variant + case-level · verified vs. 54 curated fixtures Analysis step Key output Final interpretation
TPMT / DPYD / SLCO1B1 / CYP2C19 / NUDT15

PGx Interpretation Pipeline

GitHub

A reproducible pharmacogenomics interpretation engine translating selected genomic variants into gene-specific allele/diplotype calls, functional phenotypes, and guideline-linked drug recommendations — built around explicit uncertainty rather than resolving genuinely ambiguous phasing by assumption. Five genes were chosen deliberately because they don't share one interpretation model: linked-variant dosage tables, activity-score summation, transporter biology, and direct multi-locus resolution. Deliberately scopes out CYP2D6 (structural-variant calling, unsolved even by specialist tools) rather than quietly shipping an incomplete answer. A standalone companion to the ACMG/AMP variant classifier below — same engineering discipline, applied to drug response instead of disease causation.

5 genes implemented end-to-end
233 automated tests, two independent runners
38 real GeT-RM (CDC) reference samples validated
5 report formats (JSON/TSV/HTML/MD/DOCX)
7 drug recommendations across 5 genes (Tier 2 evidence)
5-state explicit uncertainty model
CLI + Nextflow DSL2 batch entry points
Python stdlib-only dataclasses Nextflow DSL2 PharmVar CPIC/ClinPGx python-docx
View Workflow Diagram
PGx Interpretation Pipeline Workflow VCF input normalized, then five gene-specific diplotype/phenotype modules run in parallel, converging through a schema validation gate and versioned Tier 2 evidence lookup into a shared report-assembly layer, exposed via both a CLI and a Nextflow DSL2 batch pipeline. PGx Interpretation Pipeline 5 genes · 233 tests, two runners · 38 real GeT-RM samples validated · stdlib-only Python VCF (selected pharmacogene loci) TPMT · DPYD · SLCO1B1 · CYP2C19 · NUDT15 regions Variant Extraction & Normalization normalize.py · phased/unphased GT · multi-allelic ALT TPMT Linked 2-variant dosage table Phase ambiguity reported explicitly, not guessed DPYD Activity-score summation, 4 independent loci Different phenotype model SLCO1B1 Direct diplotype lookup Transporter, not an enzyme — zero schema change required CYP2C19 3 independent loci, resolved directly (not declined) Evidence-backed judgment call NUDT15 Single locus, no phasing Simplest case; first joint two-gene recommendation schema.validate() — Structural Uncertainty Gate Enforces 5 confidence states: SUPPORTED · AMBIGUOUS · INSUFFICIENT_DATA · UNSUPPORTED_ALLELE · UNRESOLVED Tier 2 Evidence Lookup — ClinPGx/CPIC (cached, versioned) 7 drug pairings across 5 genes · compound TPMT+NUDT15 · CYP2C19 dual-drug Report Assembly JSON · TSV · HTML · Markdown · DOCX — one shared, tested intermediate representation CLI — cli.py report single-sample entry point Nextflow DSL2 — main.nf fans CLI across a samplesheet Analysis step Uncertainty gate / key output Orchestration entry point
COLO829 / COLO829BL

Somatic Genomics Pipeline

GitHub

Tumour-normal somatic variant calling pipeline built around GATK Mutect2, benchmarked against a real published truth set for the COLO829/COLO829BL melanoma reference cell line pair. Reads are QC'd and aligned before contamination estimation and an interval-sharded Mutect2 call set — rebuilt as a genome-scale scatter/gather architecture after an unsharded run hit a real memory ceiling — is benchmarked with som.py and extended with CNVkit copy-number calling. A final module annotates PASS calls with SnpEff and queries CIViC's live clinical-evidence API, carrying a called variant through to real oncology evidence rather than stopping at a filtered VCF. Run against real ENA-extracted sequencing reads, the pipeline independently recovered three of COLO829's own documented driver mutations — BRAF V600E, a CDKN2A frameshift, and a UV-signature TERT promoter mutation — with zero false positives against the benchmark truth set.

8/8 PASS calls confirmed true positive (zero false positives, NYGC truth set)
3 real driver mutations independently matched: BRAF V600E · CDKN2A · TERT
132 real CIViC evidence rows retrieved live for BRAF V600E
Interval-sharded Mutect2 — genome-scale scatter/gather rebuilt after a real OOM
Real ENA-extracted tumour/normal reads (not synthetic), HiSeq X Ten WGS
som.py benchmarking + CNVkit copy-number calling alongside variant calling
Two Nextflow profiles: 8-gene melanoma driver panel + full genome
Nextflow DSL2 GATK Mutect2 CNVkit SnpEff CIViC som.py BWA-MEM2 Docker Python
View Workflow Diagram
COLO829 Somatic Genomics Pipeline Workflow Tumour-normal reads through QC, alignment and dedup, then contamination estimation and interval-sharded Mutect2 calling to a PASS-call gate, forking into copy-number calling, truth-set benchmarking, and SnpEff/CIViC clinical-evidence annotation, converging on the real-data milestone. COLO829 Somatic Genomics Pipeline BRAF · CDKN2A · TERT drivers confirmed on real data · Nextflow DSL2 · Docker Real tumour + normal FASTQ (ENA) COLO829 / COLO829BL · HiSeq X Ten WGS, gene-panel extracted FASTQC_MULTIQC raw QC — tumour + normal BWA_MEM2_ALIGN + SORT_MARKDUP dedup → analysis-ready BAMs GET_PILEUP_SUMMARIES + CALCULATE_CONTAMINATION cross-sample contamination estimate SPLIT_INTERVALS → N × MUTECT2 → MERGE interval scatter/gather, rebuilt after single-call OOM FILTER_MUTECT_CALLS — 8 PASS calls of 276 candidate records · gene-panel scope (dev profile) ① Copy-Number Calling ② Truth-Set Benchmarking ③ Oncology-Evidence Annotation CNVKIT_BATCH → CNVKIT_CALL tumour vs. normal coverage ratio Genome-wide CN segments panel-only coverage — disclosed scope limit SOMPY_BENCHMARK vs. NYGC published truth set 8/8 PASS calls = true positive 0 false positives · precision 1.0, all classes SNPEFF_ANNOTATE → CIVIC_ANNOTATE protein HGVS.p → live CIViC query 132 CIViC evidence rows BRAF V600E · real curated clinical evidence Real-data milestone BRAF V600E · CDKN2A · TERT — cross-checked against Cellosaurus & OncoKB Analysis step Key output Real-data milestone

Technical Capabilities

Pipeline & Workflow
Nextflow DSL2 nf-core bash Git
Containers & Environments
Docker Singularity conda
Metagenomics & Microbiome Analysis
QIIME 2 DADA2 nf-core/ampliseq SILVA Kraken2 Sylph KMA ResFinder FastQC MultiQC BBDuk
AMR & Typing
AMRFinderPlus MLST Kleborate emmtyper
Phylogenomics
Parsnp Snippy Gubbins IQ-TREE QUAST MAFFT
Clinical Genomics
FastQC fastp BWA-MEM2 GATK DeepVariant VEP gnomAD hap.py ACMG/AMP GIAB benchmarking PharmVar CPIC/ClinPGx GeT-RM benchmarking
Cancer / Somatic Genomics
GATK Mutect2 CNVkit SnpEff CIViC som.py
Data & Visualisation
Python Dash matplotlib pandas Biopython
Comparative Genomics & Assay Design
Inclusivity analysis Exclusivity analysis Primer3 Panaroo NCBI BLAST fastANI

Get in Touch

If you have a diagnostics, genomics or bioinformatics problem you would like to discuss, or want to talk through any of the case studies, I'd be glad to hear from you. A short description of the question and your timeline is enough to start.

How I handle your information: Privacy Notice