Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Correspondence
  • Published:

Computational correction of index switching in multiplexed sequencing libraries

Nature Methods volume 15, pages 305–307 (2018)Cite this article

Subjects

To the Editor:

Contemporary sequencers allow for large-scale multiplexing of sequencing libraries. In the single-cell field, hundreds or thousands of libraries are typically sequenced in the same lane of an Illumina sequencer1,2. It is assumed that the integrity of each library depends on the experimental conditions before sequencing, but cross-library contamination can also occur during sequencing3. Newer Illumina sequencers that use exclusion amplification (e.g., HiSeq 3000, HiSeq 4000 and NovaSeq) seem especially vulnerable to index switching caused by free index primers, which generate 2–10% contaminated reads4,5,6. Libraries for which a single swapping event can lead to the erroneous assignment of sequence reads to cells (or samples) are most affected and include single-cell libraries multiplexed by combinatorial barcoding with i5 and i7 Nextera primers. We describe a strategy to estimate the fraction of affected counts and introduce a computational correction procedure for affected data. We observed that this correction removed most cross-contamination signal and that the corrected expression data no longer clustered by index or plate, leading to the rescue of previously sequenced libraries.

This is a preview of subscription content, access via your institution

Relevant articles

Open Access articles citing this article.

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

Figure 1: Computational correction of index switching.

References

  1. Segerstolpe, Å. et al. Cell Metab. 24, 593–607 (2016).

    Article  CAS  Google Scholar 

  2. Cao, J. et al. Science 357, 661–667 (2017).

    Article  CAS  Google Scholar 

  3. Illumina Inc. Effects of Index Misassignment on Multiplexing and Downstream Analysis. Publication No. 770-2017-004-C QB # 5420 (Illumina Inc., 2017).

  4. Sinha, R. et al. bioRxiv preprint at https://www.biorxiv.org/content/early/2017/04/09/125724 (2017).

    Google Scholar 

  5. Griffiths, J.A., Richard, A.C., Bach, K., Lun, A.T.L. & Marioni, J.C. bioRxiv preprint at https://www.biorxiv.org/content/early/2018/01/30/177048 (2018).

    Google Scholar 

  6. Costello, M. et al. bioRxiv preprint at https://www.biorxiv.org/content/early/2017/10/10/200790 (2017).

    Google Scholar 

Download references

Acknowledgements

This work was supported by the Swedish Research Council (grant 2017-01062 to R. Sandberg), the Bert L. and N. Kuggie Vallee Foundation (to R. Sandberg), a National Science Foundation Graduate Research Fellowship (grant DGE 1147470 to G.S.), the NIH (grant R01CA86065 to I.L.W.), the California Institute for Regenerative Medicine (grant GC1R-06673-A), to M. Snyder; Collaborative Reserch Project to I.L.W. and the Virginia and D.K. Ludwig Fund for Cancer Research (to I.L.W.)

Author information

Authors and Affiliations

  1. Department of Cell and Molecular Biology, Karolinska Institutet, Stockholm, Sweden

    Anton J M Larsson & Rickard Sandberg

  2. Department of Bioengineering, Stanford University School of Engineering, Stanford, California, USA

    Geoff Stanley

  3. Institute for Stem Cell Biology and Regenerative Medicine, Stanford University School of Medicine, Stanford, California, USA

    Rahul Sinha & Irving L Weissman

Authors
  1. Anton J M Larsson
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Geoff Stanley
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Rahul Sinha
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Irving L Weissman
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Rickard Sandberg
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Rickard Sandberg.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Integrated supplementary information

Supplementary Figure 1 Read count distributions at different thresholds of filtering.

(A) Histogram of gene read count distributions (post-correction HiSeq4000) for all 384 cells of the HSC plate, after applying read count filtering at the indicated threshold (1 to 10 reads). (B) The read count distribution of the same plate (sequenced on NextSeq) without correction or read count filtering. We noted that corrected data still had an excess of low read counts which we eliminated by using a read-count threshold. Note that using a read count threshold is not sufficient to remove cross-contamination signal (Fig. S2).

Supplementary Figure 2 Evaluation of read count threshold for index switching correction.

(A) Histogram of false positive gene expression, defined as gene expression detected on the HiSeq 4000 before (blue) and after (yellow) correction where no expression was detected in the same cell on the NextSeq 500. (B) Histogram with fraction false positive expression signals removed per gene. (C) Histogram of false negative gene expression defined as no detectable gene expression on the HiSeq 4000 before (blue) and after (yellow) correction where expression was detected in the same cell on the NextSeq 500.

Supplementary Figure 3 Corrected data rescued index-driven HSC sub-clusters.

(A) Robust PCA (rPCA) analyses of HSCs based on HiSeq 4000 sequence counts, coloured by the Nextera i5 index. The rPCA algorithm assigns the most prominent outlier observations (i.e. cells) into the last principal components, which corresponded to wells that had been amplified using the same i7 and i5 indices (n = 384 cells). (B) Loadings of cells on PC24, stratified by i5 Nextera primers (n = 384 cells, 16 stratifications) (C) Loading of cells on P25, stratified by i7 Nextera primers (n = 384 cells, 24 stratifications). Center: Median, Hinges: 1st and 3rd quartiles, Whiskers: 1.5 interquartile range (IQR) (D) rPCA clustering of HSCs (as in A) for corrected HiSeq 4000 sequence counts, colored by Nextera i5 index primer. (E-F) As in (C-D) for corrected HiSeq 4000 sequence counts.

Supplementary Figure 4 Corrected data rescued index-driven plate-associated clustering.

(A) PCA analysis of HSCs from two plates of libraries sequenced on the HiSeq 4000, colored by library plate. PC2 scores separate cells by plates (n = 384 cells per plate). (B) As in (A) for corrected transcriptome data.

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–4 and Supplementary Methods (PDF 938 kb)

Life Sciences Reporting Summary (PDF 67 kb)

Supplementary Software

unspread software for index switching correction (ZIP 6 kb)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Larsson, A., Stanley, G., Sinha, R. et al. Computational correction of index switching in multiplexed sequencing libraries. Nat Methods 15, 305–307 (2018). https://doi.org/10.1038/nmeth.4666

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nmeth.4666

This article is cited by

Search

Advanced search

Quick links

Nature Briefing: Translational Research

Sign up for the Nature Briefing: Translational Research newsletter — top stories in biotechnology, drug discovery and pharma.

Get what matters in translational research, free to your inbox weekly. Sign up for Nature Briefing: Translational Research