Report Quality Alerts
At the beginning of the DRAGEN report, Quality Alerts highlight metrics that differ significantly from expected values for healthy tissue and high-quality samples. Depending on your experimental design or sample specifics, these alerts may be safely ignored, or they may signal issues in the experiment or analysis.
Quick Reference Table
SBC Reads Mapping
Small tissue slices, over-permeabilization, wrong Map file, incorrect images, or mismatched metadata
Payload Sequencing Quality
Poor sequencing run quality
Reads Removed by Trimming
Poor quality sample, poor library prep
PolyA Trimming
Poor quality sample
Quality Trimming
Poor sequencing quality
Unique Alignments
Poor quality sample, wrong genome reference
Reads Aligned to Genes
Poor quality sample, wrong genome reference
rRNA Percentage
High ribosomal RNA content
mtRNA Percentage
High mitochondrial RNA content
Cell Segmentation Probability
Blurry or out-of-focus images, section thickness
Transcripts Inside Cells
Poor registration, poor segmentation, high diffusion
MT Filter Loss
Default threshold not ideal for sample
UMI Filter Loss
Default threshold not ideal for sample
Detailed Metric Descriptions
1. SBC Reads Mapping (Run Level)
What it measures: Percentage of sequencing reads that successfully map to Spatial Barcode (SBC) sequences.
Thresholds:
NOTE: <60%
WARNING: <10%
What causes this alert:
NOTE (<60%): This can be due to small tissue slices, high diffusion from over-permeabilization, or other technical issues.
WARNING (<10%): This can indicate the wrong Map file, incorrect images, or mismatched metadata were used for this analysis.
What to do:
NOTE: Review your tissue section size and permeabilization protocol. Small tissue samples naturally have fewer reads mapping to SBCs. If over-permeabilization is suspected, adjust permeabilization time in future experiments.
WARNING: Verify that the correct slideID, images, and sample metadata were used for this analysis run.
2. Payload Sequencing Quality (Sample Level)
What it measures: Mean quality score of the payload (insert) sequences after trimming.
Thresholds:
NOTE: <35
WARNING: <30
What causes this alert: This can indicate the sequencing run was poor quality. Check FastQC/MultiQC reports to examine sequencing quality further.
What to do:
Review the FastQC and MultiQC reports included in your results.
If WARNING, the data quality may be too poor for reliable analysis. Consider re-sequencing the library.
If this affects multiple samples from the same sequencing run, there may have been an issue with the sequencer.
3. Reads Removed by Trimming (Sample Level)
What it measures: Percentage of reads that were completely removed during the trimming process.
Thresholds:
NOTE: >20%
WARNING: >40%
What causes this alert: This can indicate a poor-quality sample and/or poor library preparation.
What to do:
NOTE: Review your library preparation protocol for potential issues. The sample may still be usable, but with reduced data.
WARNING: The high removal rate suggests significant quality issues. Review sample handling, storage, and library prep steps.
Consider whether the sample was degraded prior to library preparation.
4. PolyA Trimming (Sample Level)
What it measures: Percentage of reads that required trimming of polyA tails.
Thresholds:
NOTE: >20%
WARNING: >50%
What causes this alert: This can indicate a poor-quality sample.
What to do:
High polyA content may indicate RNA degradation.
Review sample collection, storage, and handling procedures.
Evaluate RNA integrity if available (RIN scores).
5. Quality Trimming (Sample Level)
What it measures: Percentage of reads that required trimming due to low base quality scores.
Thresholds:
NOTE: >10%
WARNING: >20%
What causes this alert: This can indicate poor sequencing quality.
What to do:
Review sequencing metrics and quality reports.
If multiple samples show this alert, there may have been a sequencer issue.
Check for any noted issues during the sequencing run.
6. Unique Alignments (Sample Level)
What it measures: Percentage of reads that align uniquely to a single location in the genome.
Thresholds:
NOTE: <70%
WARNING: <40%
What causes this alert:
NOTE (<70%): This can indicate a poor-quality sample.
WARNING (<40%): This can indicate that the wrong genome reference was used for the analysis.
What to do:
NOTE: Review sample quality metrics. Lower unique alignment rates can occur with degraded samples or high levels of ribosomal or mitochondrial RNA.
WARNING: First, verify that the correct genome reference (species and version) was selected for the analysis. If incorrect, re-run the analysis with the correct reference.
7. Reads Aligned to Genes (Sample Level)
What it measures: Percentage of mapped reads (including multimapped reads when enabled) that align to annotated gene regions.
Thresholds:
NOTE: <70%
WARNING: <50%
What causes this alert:
NOTE (<70%): This can indicate a poor-quality sample.
WARNING (<50%): This can indicate that the wrong genome reference was used for the analysis.
What to do:
NOTE: Low gene alignment can result from degraded RNA or high levels of non-coding RNA. Review other quality metrics.
WARNING: Verify that the correct genome reference and gene annotation were used for the analysis.
8. rRNA Percentage (Sample Level)
What it measures: Percentage of UMIs (unique molecular identifiers) that correspond to ribosomal RNA.
Thresholds:
NOTE: >20%
No WARNING threshold
What causes this alert: This sample has a high level of rRNA. This may be expected based on the specific sample or could merit further investigation.
What to do:
High rRNA content is common in many sample types and may not indicate a problem.
This is primarily informational; evaluate based on your specific experimental design and sample type.
9. mtRNA Percentage (Sample Level)
What it measures: Percentage of UMIs that correspond to mitochondrial RNA.
Thresholds:
NOTE: >40%
No WARNING threshold
What causes this alert: This sample has a high level of mtRNA. This may be expected based on the specific sample or could merit further investigation.
What to do:
High mitochondrial RNA content is expected in certain tissue types (heart, muscle, kidney, liver, brain).
It can also indicate cellular stress or dying/damaged cells.
This is primarily informational; consider your biological context.
If unexpected for your tissue type, it may warrant investigation into sample handling.
10. Cell Segmentation Probability (Sample Level)
What it measures: Median probability score from the cell segmentation algorithm, indicating confidence in cell boundary detection.
Thresholds:
NOTE: <75%
WARNING: <50%
What causes this alert: This can indicate blurry or out-of-focus images, and/or may be related to the thickness of the section.
What to do:
Review the original microscopy images for focus and clarity.
Check if imaging conditions were optimal.
Evaluate whether section thickness may be affecting image quality.
If images are poor quality, consider whether re-imaging is possible and necessary for your analysis.
Poor segmentation will reduce the accuracy of cell-level analysis.
For WARNING alerts, cell boundaries may be unreliable; consider using bin-based analysis instead if available.
11. Transcripts Inside Cells (Sample Level)
What it measures: Percentage of detected transcripts that fall within segmented cell boundaries.
Thresholds:
NOTE: <50%
WARNING: <35%
What causes this alert: This can indicate poor registration or poor cell segmentation and/or high diffusion.
What to do:
Check the quality of image registration (alignment between imaging and sequencing data).
Review cell segmentation quality (see Cell Segmentation Probability above).
Evaluate whether over-permeabilization may have caused transcript diffusion outside cells.
Low values will affect the accuracy of cell-level gene expression profiles.
Review tissue handling and permeabilization protocols for future experiments.
12. MT Filter Loss (Sample Level)
What it measures: Percentage of cells or bins that were filtered out during clustering due to high mitochondrial RNA content.
Thresholds:
NOTE: >25%
WARNING: >50%
What causes this alert:
NOTE (>25%): A high percentage of cells were filtered out during clustering because of high mtRNA. The default threshold may not be ideal for your sample.
WARNING (>50%): A very high percentage of cells were filtered out during clustering because of high mtRNA. The default threshold may not be ideal for your sample.
What to do:
This alert indicates that the default MT filtering threshold (used to remove low-quality cells) may be too stringent for your sample type.
High MT content is normal for certain tissues (heart, muscle, kidney, liver, brain).
Consider adjusting the MT filtering threshold in your analysis parameters using analysis_config.json or directly in ICM.
13. UMI Filter Loss (Sample Level)
What it measures: Percentage of cells or bins that were filtered out during clustering due to low UMI (molecule) counts.
Thresholds:
NOTE: >25%
WARNING: >50%
What causes this alert:
NOTE (>25%): A high percentage of cells were filtered out during clustering because of low molecule counts. The default threshold may not be ideal for your sample.
WARNING (>50%): A very high percentage of cells were filtered out during clustering because of low molecule counts. The default threshold may not be ideal for your sample.
What to do:
This alert indicates that the default UMI filtering threshold (used to remove low-quality cells and bins) may not be optimal for your sample.
Consider whether your sample naturally has lower transcript counts (for example, certain cell types or small cells).
Evaluate adjusting the UMI filtering threshold in your analysis parameters using analysis_config.json or directly in ICM.
Understanding Alert Levels
NOTE
Indicates metrics outside the typical range for high-quality samples. These may be acceptable depending on your experimental design and sample characteristics. Review the specific alert and consider your biological context.
WARNING
Indicates metrics severely outside expected ranges, often suggesting technical issues, incorrect analysis parameters, or very poor sample quality. These generally require action such as verifying analysis parameters, reviewing sample quality, or potentially re-running parts of the analysis.
Note: Threshold values and messages in this guide are current as of April 30, 2026. Default thresholds may be updated based on ongoing data analysis and customer feedback.
Last updated
Was this helpful?