Pipeline Update: Human Read Filtering

Jump to Section:

• Background
• Validation

What's New?

As of February 2, 2024, CZ ID has been updated to include a new reference for human read filtering. This update improves the removal of human reads from uploaded samples by mapping reads to two human genome assemblies, namely HG38 and T2T-CHM13.

To take advantage of the updated human reference including T2T-CHM13, make sure to upload samples to projects created on or after February 2, 2024. Samples uploaded to projects created before February 2, 2024 will be processed using the HG38 human reference alone.

---

Background

Human read filtering during data preprocessing is an important step to reduce the number of contaminant human sequences in next generation sequencing (NGS) samples and to protect personally identifiable information. CZ ID removes human reads through subtractive alignment, where reads are mapped to a reference human genome and then mapped reads are removed from the dataset. Therefore, the choice of the human reference genome(s) has a direct impact on the accuracy of results. In January 2024, the CZ ID team updated the human reference genome to include the most complete human genome assembly available to date to improve the detection and removal of human reads from uploaded datasets. 

Prior to February 2024, CZ ID only implemented the human genome build 38 assembly (GRCh38 or HG38; accession GCF_000001405.26) for human read filtering. After February 2, 2024, the new reference used for human read filtering includes HG38 and the latest assembly from the Telomere-to-Telomore (T2T) Consortium completed in 2022 (T2T-CHM13v2.0; accession GCF_009914755.1). The T2T-CHM13 genome increased the number of known human genes and added or corrected 238 Mbp of sequence, primarily comprising centromeric satellites, non-satellite segmental duplications, and rDNAs (Nurk et al. 2022). 

---

Validation

Our team validated that the new human reference containing HG38 and T2T-CHM13 genomes improved human read filtering. To do this, our team used two datasets from the HG002 human cell line sequenced using Illumina (Human Pangenome Reference Consortium) and Nanopore (Kolmogorov et al. 2023) to explore human read filtering through CZ ID’s metagenomic (mNGS) pipelines. These datasets should only contain reads representing human and Epstein-Barr virus (used to transform the cell lines). Consistent with previous findings (Rhie et al. 2023; Wang and Xing 2023), the updated reference increased the number of human reads removed from the HG002 datasets when compared to the reference containing HG38 alone (Table 1).

Table 1. Number of human reads removed from HG002 human cell line datasets after mapping reads to HG38 or HG38 and T2T-CHM13 reference genomes.

CZ ID Pipeline	No. Human Reads Removed
	HG38	HG38+T2T-CHM13	Read Difference (%)
mNGS Illumina	68,926,530	69,123,514	196,984 (0.26%)
mNGS Nanopore	309,013	309,043	30 (0.01%)

Updating the reference for human read filtering had a larger effect on the short-read (Illumina) than the long-read (Nanopore) mNGS pipeline. The increased removal of human reads prior to analysis reduced the number of false positives in mNGS sample reports. The reduction in false positives is likely due to contaminating human DNA found in datasets representing other species in NCBI databases (Wang and Xing 2023).

---

References

Kolmogorov M, Billingsley KJ, Mastoras M, Meredith M, Monlong J, Lorig-Roach R, et al. Scalable Nanopore sequencing of human genomes provides a comprehensive view of haplotype-resolved variation and methylation. bioRxiv. 2023. DOI: 10.1101/2023.01.12.523790

Nurk S, Koren S, Rhie A, Rautiainen M, Bzikadze AV, Mikheenko A, et al. The complete sequence of a human genome. Science. 2022;376: 44–53. DOI: 10.1126/science.abj6987

Rhie A, Nurk S, Cechova M, Hoyt SJ, Taylor DJ, Altemose N, et al. The complete sequence of a human Y chromosome. Nature. 2023;621: 344–354. DOI: 10.1038/s41586-023-06457-y

Wang L, Xing G. Telomere-to-Telomere Assembly Improves Host Reads Removal in Metagenomic High-Throughput Sequencing of Human Samples. bioRxiv. 2023. DOI: 10.1101/2023.05.05.539517