Related Experiment Video
Updated: Jul 15, 2026
07:10
Depletion of Mouse Cells from Human Tumor Xenografts Significantly Improves Downstream Analysis of Target Cells
Published on: July 29, 2016
HumanFilt: a multi-reference host depletion pipeline improves Fusobacterium detection accuracy in tumor WGS data sets
Mariia Frolova1, Barry Maguire1,2, Heiko Duessmann1
1Department of Physiology and Medical Physics and RCSI Centre for Systems Medicine, Royal College of Surgeons in Ireland, Dublin, Ireland.
Msystems
|July 14, 2026
Summary
HumanFilt, an open-source tool, effectively removes human DNA contamination from whole-genome sequencing data, improving tumor microbiome analysis. This method preserves crucial microbial signals like Fusobacterium, essential for understanding tumor-bacteria interactions.
Area of Science:
- Microbiome research
- Genomics
- Bioinformatics
Background:
- Whole-genome sequencing (WGS) for tumor-associated microbiomes is gaining traction.
- Detecting microbial signals is challenging due to host DNA contamination and artifacts.
- Efficient removal of human reads is critical for accurate microbiome analysis.
Purpose of the Study:
- Introduce HumanFilt, an open-source tool for rigorous removal of human-derived sequences from WGS data.
- Apply HumanFilt to re-analyze WGS data from rectal adenocarcinoma cases.
- Assess the impact of host-read removal on microbial signal detection and reliability.
Main Methods:
- Developed HumanFilt, integrating k-mer classification (Kraken2), trimming (Trim Galore), vector filtering (BBDuk/UniVec_Core), and deduplication (FastUniq).
- Employed a multi-aligner, multi-reference approach (BWA-MEM, Bowtie2, Minimap2) for residual host sequence removal.
- Utilized taxonomic profiling (PathSeq, MetaPhlAn) and cross-validated with immunofluorescence.
Main Results:
- HumanFilt eliminated over 99.9% of human-derived reads, removing millions of residual sequences per sample.
- Reproducible enrichment of Fusobacterium species in tumor vs. normal tissues was preserved post-filtering.
- HumanFilt demonstrated high specificity, retaining over 99.7% of true Fusobacterium signals, with a trade-off in ambiguous signal preservation compared to other tools.
Conclusions:
- Rigorous host-read depletion using HumanFilt reduces artificial microbial signals in normal samples while preserving tumor-associated microbes like Fusobacterium.
- The tool provides a more reliable microbial profile for studying tumor-bacteria interactions.
- Findings highlight the importance of stringent filtering and the trade-offs in host DNA removal for microbiome studies.
