Qualimap

About Qualimap

Qualimap is a platform-independent application for quality control of sequencing alignment data, providing gene body coverage profiling, read origin classification, and bias metrics for RNA-Seq experiments.

Documentation | Source code

What is Qualimap?

Qualimap is a platform-independent application for quality control of sequencing alignment data. RustQC reimplements the rnaseq mode of Qualimap, producing identical output that is directly parseable by MultiQC.

The Qualimap analysis runs automatically as part of rustqc rna in the same single-pass BAM scan as all other analyses — no extra BAM pass is needed.

What it provides

RustQC’s Qualimap module computes:

• Gene body coverage profiles: coverage distribution along normalized transcript positions from 5’ to 3’, revealing systematic biases such as 3’ degradation or incomplete reverse transcription
• 5’/3’ bias metrics: quantitative measures of coverage uniformity across gene bodies, used for sample QC and batch-effect detection
• Read origin classification: counts reads as exonic, intronic, intergenic, or overlapping multiple genes’ exons
• Strand-specificity estimation: validates whether the observed strand distribution matches the expected library protocol
• Splice junction motif analysis: counts canonical and non-canonical donor/acceptor junction motifs across all uniquely-mapped reads
• Alignment statistics: primary/secondary alignments, multi-mappers, proper pairs, and supplementary read counts

No name-sorted BAM required

The original Qualimap requires name-sorted BAM files as input, which typically means adding an extra samtools sort -n step to your pipeline. RustQC works directly with coordinate-sorted BAM files (the standard output of most aligners), so you can skip this sorting step entirely.

Output files

All Qualimap output files are written to a qualimap/ subdirectory under the output directory, matching the directory structure that MultiQC expects.

• Directoryqualimap/

  • qualimapReport.html Self-contained HTML report
  • rnaseq_qc_results.txt Qualimap-compatible QC report
  • Directoryraw_data_qualimapReport/

    • coverage_profile_along_genes_(total).txt Coverage profile (all genes)
    • coverage_profile_along_genes_(high).txt Coverage profile (highly expressed genes)
    • coverage_profile_along_genes_(low).txt Coverage profile (lowly expressed genes)
  • Directoryimages_qualimapReport/

    • Coverage Profile Along Genes (Total).png Gene body coverage chart
    • Coverage Profile Along Genes (High).png Coverage chart (highly expressed genes)
    • Coverage Profile Along Genes (Low).png Coverage chart (lowly expressed genes)
    • Transcript coverage histogram.png Coverage depth histogram (0-50X)
    • Reads Genomic Origin.png Exonic/intronic/intergenic pie chart
    • Junction Analysis.png Known/novel junction pie chart

QC results

File: rnaseq_qc_results.txt

The main QC report, formatted identically to Qualimap’s rnaseq output. This file is the primary input for MultiQC’s Qualimap rnaseq module. It contains four sections:

Input

>>>>>>> Input
     bam file = sample

Reads alignment

Alignment-level statistics for all primary reads:

>>>>>>> Reads alignment
     reads aligned (left/right) = 86,909,417 / 86,888,892
     read pairs aligned = 86,778,882
     total alignments = 189,114,475
     secondary alignments = 15,316,166
     non-unique alignments = 26,451,119
     aligned to genes = 122,500,822
     ambiguous alignments = 2,422,409
     no feature assigned = 37,740,125
     not aligned = 12,490,977
     SSP estimation (fwd/rev) = 0.04 / 0.96

Field	Description
reads aligned	Total primary alignments (excluding secondary, supplementary, QC-fail)
total alignments	All mapped reads (primary + secondary)
secondary alignments	Alignments flagged as secondary (0x100)
non-unique alignments	Multi-mappers (NH tag > 1)
aligned to genes	Reads assigned to exactly one gene (exonic)
ambiguous alignments	Reads enclosed by multiple genes
no feature assigned	Reads with no enclosing gene annotation
not aligned	Unmapped reads encountered in the BAM
SSP estimation	Strand-specificity proportions (forward / reverse)

Reads genomic origin

Per-base classification of read origin, counting each read’s aligned bases against the annotation:

>>>>>>> Reads genomic origin
     exonic = 122,500,822 (76.45%)
     intronic = 29,504,572 (18.37%)
     intergenic = 8,344,022 (5.19%)
     overlapping exon = 6,754,726 (4.20%)

Field	Description
exonic	Reads fully enclosed by a single gene’s exons
intronic	Reads overlapping introns but not exons
intergenic	Reads in regions with no gene annotation
overlapping exon	Reads partially overlapping exons (not fully enclosed)

Transcript coverage profile

Coverage bias metrics computed from the gene body coverage profile:

>>>>>>> Transcript coverage profile
     5' bias = 0.87
     3' bias = 0.77
     5'-3' bias = 1.11

Metric	Description
5’ bias	Ratio of median coverage in the 5’ region (first 100bp) to the median of the middle 80%
3’ bias	Ratio of median coverage in the 3’ region (last 100bp) to the median of the middle 80%
5’-3’ bias	Ratio of 5’ median to 3’ median

Bias values close to 1.0 indicate uniform coverage. Values significantly above 1.0 for 5’ bias or below 1.0 for 3’ bias suggest degradation or protocol issues.

Coverage profiles

Files: coverage_profile_along_genes_(total|high|low).txt

Tab-separated files with 100 rows, one per percentile bin (0-99), showing the normalized coverage depth at each position along the gene body:

Column	Description
Position	Percentile position along gene body (0.0 to 99.0)
Coverage	Normalized coverage depth at this position

Three tiers are produced:

Tier	File	Description
Total	coverage_profile_along_genes_(total).txt	All expressed genes
High	coverage_profile_along_genes_(high).txt	Top 500 genes by mean coverage
Low	coverage_profile_along_genes_(low).txt	Bottom 500 genes by mean coverage

The total profile is the one parsed by MultiQC for the gene body coverage plot.

Qualimap chart images

Chart images are generated in images_qualimapReport/, matching the plots produced by Qualimap’s HTML report:

Coverage Profile Along Genes (Total)

Qualimap (Java)

RustQC

Coverage Profile Along Genes (High)

Qualimap (Java)

RustQC

Coverage Profile Along Genes (Low)

Qualimap (Java)

RustQC

Transcript coverage histogram

Qualimap (Java)

RustQC

Reads Genomic Origin

Qualimap (Java)

RustQC

Junction Analysis

Qualimap (Java)

RustQC

HTML report

File: qualimapReport.html

A self-contained HTML report matching Qualimap’s layout and styling, including all six chart images, summary tables, and a sidebar table of contents. The report uses embedded CSS and base64-encoded assets so it can be viewed in any browser without external dependencies.

View an example HTML report

Interpreting results

Gene body coverage shape

Pattern	Interpretation
Flat/uniform	Good library quality — even coverage across transcripts
3’ bias (rising right)	RNA degradation or oligo-dT priming bias
5’ bias (rising left)	Incomplete reverse transcription or random-priming protocol
Both-end dropout	Short fragment library or size-selection issue

Bias metrics

Metric	Good	Problematic
5’-3’ bias	0.8 - 1.2	< 0.5 or > 2.0
5’ bias	0.8 - 1.2	> 2.0 (enrichment) or < 0.5 (depletion)
3’ bias	0.8 - 1.2	> 2.0 (enrichment) or < 0.5 (depletion)

Strand specificity

For strand-specific libraries, one of the SSP values should be close to 1.0 and the other close to 0.0. Common patterns:

Library type	Expected fwd	Expected rev
Unstranded	~0.5	~0.5
dUTP / Illumina TruSeq Stranded	~0.02-0.05	~0.95-0.98
Ligation (e.g. SMARTer)	~0.95-0.98	~0.02-0.05

Benchmarks

RustQC’s Qualimap analysis runs in the same single-pass BAM scan as all other tools, eliminating the separate name-sort and Qualimap run required by the original Java implementation.

Performance

Large dataset (~186M reads, GM12878)

Tool	Runtime	Peak RSS	Notes
samtools sort -n	12m 16s	5.2 GB	Name-sort required by Qualimap Java (not needed by RustQC)
Qualimap rnaseq (Java)	49m 58s	4.9 GB	Separate BAM pass, single-threaded, requires name-sorted input
Total upstream pipeline	~1h 2m 14s		Name-sort + Qualimap combined
RustQC (all tools)	14m 54s	11.4 GB	Single pass, all analyses combined, coordinate-sorted input

Small dataset (~52K reads, chr6)

Tool	Runtime	Peak RSS	Notes
Qualimap rnaseq (Java)	5.0s	704.6 MB	Separate BAM pass, requires name-sorted input
RustQC (all tools)	25.9s	182.1 MB	Single pass, all analyses combined

Note: RustQC runtime shown is for all tools combined in a single pass. See Benchmark Details for a full breakdown.

On the small dataset, RustQC uses 182 MB of memory vs Qualimap’s 705 MB.

Output comparison

RustQC produces output effectively identical to Qualimap Java. Both tools were run on the same BAM file with matching parameters (--stranded reverse / strand-specific-reverse, same GENCODE GTF). Results below are from the AWS benchmark run (2026-03-09).

Deterministic output

Qualimap uses HashMap internally, whose iteration order is not guaranteed across JVM versions. This means bias metrics (especially 5’-3’ bias) can vary slightly between runs or environments in the Java version. RustQC avoids this by using a stable tie-breaking rule when multiple transcripts of the same gene have identical mean coverage.

Small dataset (~52K reads, chr6)

Reads alignment

Metric	Qualimap (Java)	RustQC	Match?
reads aligned (left)	24,797	24,797	Yes
reads aligned (right)	24,776	24,776	Yes
read pairs aligned	24,773	24,773	Yes
total alignments	52,839	52,839	Yes
secondary alignments	3,266	3,266	Yes
non-unique alignments	5,496	5,496	Yes
aligned to genes	31,654	31,654	Yes
ambiguous alignments	1,054	1,054	Yes
no feature assigned	14,635	14,635	Yes
not aligned	0	0	Yes

Reads genomic origin

Metric	Qualimap (Java)	RustQC	Match?
exonic fraction	68.38%	68.38%	Yes
intronic fraction	29.57%	29.57%	Yes
intergenic fraction	2.04%	2.04%	Yes

Transcript coverage profile

Metric	Qualimap (Java)	RustQC	Match?
5’ bias	0.71	0.71	Yes
3’ bias	0.57	0.57	Yes
5’-3’ bias	1.30	1.19	Minor diff

The 5’-3’ bias shows a minor difference (1.30 vs 1.19) on this small dataset. All other values are identical.

Large dataset (~186M reads, GM12878)

Reads alignment

Metric	Qualimap (Java)	RustQC	Match?
reads aligned (left)	86,909,417	86,909,417	Yes
reads aligned (right)	86,888,892	86,888,892	Yes
read pairs aligned	86,778,882	86,778,882	Yes
total alignments	189,114,475	189,114,475	Yes
secondary alignments	15,316,166	15,316,166	Yes
non-unique alignments	26,451,119	26,451,119	Yes
aligned to genes	122,500,822	122,500,822	Yes
ambiguous alignments	2,422,409	2,422,409	Yes
no feature assigned	37,740,125	37,740,125	Yes
not aligned	12,490,977	12,490,977	Yes

Reads genomic origin

Metric	Qualimap (Java)	RustQC	Match?
exonic fraction	76.45%	76.45%	Yes
intronic fraction	18.37%	18.37%	Yes
intergenic fraction	5.19%	5.19%	Yes
overlapping exon	4.20%	4.20%	Yes

Transcript coverage profile

Metric	Qualimap (Java)	RustQC	Match?
5’ bias	0.87	0.87	Yes
3’ bias	0.77	0.77	Yes
5’-3’ bias	1.11	1.11	Yes

Junction analysis

Metric	Qualimap (Java)	RustQC	Match?
reads at junctions	56,119,719	56,119,719	Yes

Junction motif type percentages match with rounding differences less than 0.05 percentage points. The underlying read counts are identical.

HTML reports

Both tools produce an HTML report with embedded charts and summary tables. View the full reports from the large benchmark run:

• Qualimap (Java) HTML report

• RustQC HTML report

Note that RustQC base64-encodes all assets within the HTML file, so it can be shared stand-alone.

Configuration

Gene body coverage can be enabled or disabled in the config. See the Configuration page for details.

References

• Qualimap: Garcia-Alcalde F, Okonechnikov K, Carbonell J, et al. Qualimap: evaluating next-generation sequencing alignment data. Bioinformatics. 2012;28(20):2678-2679. Qualimap website