# TruPath Outputs
The proximity mode enabled DRAGEN Germline pipeline for use with the Illumina TruPath Genome prep produces a layered set of outputs that begin with **proximity‑aware templates and alignments,** expand into **long‑range phasing**, and culminate in **phased small variant calls, haplotype‑resolved SVs, small variants in paralogous regions (MRJD), STR expansion calls, and colocation‑validated break ends,** all backed by extensive QC and reporting. Each algorithm consumes the same underlying proximity signal but exposes results through **standard genomics artifacts (BAM, VCF, CSV, JSON, Cooler)**, making the workflow powerful yet interoperable.
For more information on the DRAGEN algorithms, features, and outputs supporting Illumina TruPath Genome prep, please navigate to the [DRAGEN User Guide linked here.](https://help.dragen.illumina.com/dragen-v4.5-trupath)
#### 1. Proximity Linking Model & Template Reconstruction
* A **non‑linear proximity linking model** is fit per read group using flow cell (X,Y) distance and genomic distance to reconstruct long DNA templates. This is the foundational signal reused by each downstream algorithm.
* Key Output Files:
* **Proximity‑aware BAM/CRAM**
* Reads tagged with **`BX:Z`** (template ID)
* **Template metrics CSVs** (WGS + QC regions):
* `._template_subpairs.csv`
* `._template_gdist.csv`
* `._template_xdist.csv`
* `._template_ydist.csv`
* `._template_thresholds.csv`
* **Link metrics CSVs**:
* `._proximity_gdist.csv`
* `._proximity_xdist.csv`
* `._proximity_ydist.csv`
#### 2. Proximity-Aware Mapping & Alignment
* Uses proximity link probabilities as an additional alignment score to resolve ambiguous mappings that standard short-read alignment cannot resolve.
* Key Output Files:
* **Mapped BAM/CRAM**
* Improved placement in repeats/paralogs
* Carries proximity and template tags
#### 3. Read Phasing & Personalized Haplotypes
* Combines haplotype databases with long‑range proximity links to create long, confident phase blocks, enabling haplotype‑aware variant calling and downstream analyses.
* Key Output Files:
* **Phased BAM/CRAM**
* Tags: `HP` (haplotype), `PS` (phase block), `pp` (phasing confidence)
* **Personalized haplotypes TSV**
* `.personal_haplotypes.tsv.gz`
* **Phase block GTF**
* `.phase_blocks.gtf.gz`
* **Imputed variants VCF**
* `.personal.vcf.gz`
* **Phasing metrics CSV**
* `.phasing_summary_stats.csv`
#### 4. Proximity-Aware Structural Variant (SV) Calling
* Uses haplotype‑segregated assemblies and phasing‑aware machine learning (ML) models to improve SV detection and genotyping in single‑sample germline WGS. Proximity information enters indirectly through phasing.
* Key Output Files
* **SV VCF**
* Includes TruPath‑specific fields:
* `PHASEDASM`
* `ML_UPDATED`
* `MLQS`
* `ColocationSum` (when colocation filter applied)
#### 5. Multi-Region Joint Detection (MRJD)
* Performs *de novo*, haplotype‑resolved small‑variant calling in paralogous regions, estimating copy number and assigning variants to specific gene copies or haplotypes using proximity information
* Key output files
* **Primary MRJD VCF**
* `.mrjd.hard-filtered.vcf.gz`
* **MRJD JSON summary**
* `.mrjd.json`
* Copy number, region/haplotype assignments, run status
* **MRJD phased BAM**
* `.mrjd.phased.bam`
* Tags: `HP` (copy), `PC` (confidence), `PS`, `BX`
* **Supporting files directory**
* `mrjd_supporting_files/`
* Multi‑column VCFs per paralog set
* Reference region alignments SAM
#### 6. STR Calling with IRR Recovery
* Recovers otherwise unmapped in‑repeat reads using proximity, enabling more accurate sizing of large STR expansions and improving genotyping in heterozygous cases.
* Key Output Files
* **STR VCF** (standard DRAGEN‑STR format)
* **BAM/CRAM with IRR tags**
* `tr` tag encoding recovered repeat motif
#### 7. Colocation Analysis & Filtering
* Summarizes long‑range genomic interactions from proximity‑linked reads and uses that signal to validate or filter SV break ends lacking molecular support
* Key Output Files:
* **Cooler file**
* Sparse colocation matrix (Hi‑C‑like)
* **SV VCF annotations**
* `NORMALIZED_COLOC_SUM`
* `ColocationSum` filter (when applied)
#### 8. Proximity‑Filtered Coverage & Reporting
* Provides QC and interpretability for proximity data quality, template structure, and phasing performance, integrated into standard DRAGEN Report
* Key Output Files:
* **Proximity coverage CSVs** (per link‑quality threshold):
* `_proximity_linkqual_coverage_metrics.csv`
* `_proximity_linkqual_hist.csv`
* `_proximity_linkqual_fine_hist.csv`
* `_proximity_linkqual_overall_mean_cov.csv`
* `_proximity_linkqual_contig_mean_cov.csv`
#### 9. DRAGEN Reports (TruPath Germline WGS)
* Dedicated *Proximity* tab with QC metrics and visualizations. See the DRAGEN Reports section of the DRAGEN User guide for additional information:
* Key Metrics:
* `Fit RMSE` - An estimate of how different the estimated probabilities can be between the parametric and non-parametric models, on the phred scale
* `Q25 Proximity Rate` - Percentage of read-pairs with at least one neighbor above Q25, on the phred scale
* `Q25 Proximity Coverage` - Average alignment coverage over genome with link-quality ≥Q25, on the phred scale
* `P75 Template Size` - The size of linked template molecules at the 75th percentile
* `NG50` - The size of the smallest phasing block required to phase 50% of the genome
* Key Plots:
* Template Genomic Span, displaying the distribution of template genomic lengths from `.wgs_template_gdist.csv`
.png?alt=media){kind=spacer}
.png?alt=media)