Tandem-genotypes: robust detection of tandem repeat expansions from long DNA reads

Abstract

Tandemly repeated DNA is highly mutable and causes at least 31 diseases, but it is hard to detect pathogenic repeat expansions genome-wide. Here, we report robust detection of human repeat expansions from careful alignments of long but error-prone (PacBio and nanopore) reads to a reference genome. Our method is robust to systematic sequencing errors, inexact repeats with fuzzy boundaries, and low sequencing coverage. By comparing to healthy controls, we prioritize pathogenic expansions within the top 10 out of 700,000 tandem repeats in whole genome sequencing data. This may help to elucidate the many genetic diseases whose causes remain unknown.

Keywords: Long-read sequencing; Nanopore; PacBio; Repeat diseases; Tandem repeat.

Fig. 1

Flow chart to predict and prioritize tandem repeat changes, using the tandem-genotypes and tandem-genotypes-join programs. Long reads are aligned to a reference genome using LAST. Dotted square: program developed in this study

Fig. 2

a–j Distribution of predicted change in repeat copy number, for nanopore reads from each of ten plasmids. Red arrows: expected copy number change. Forward (red) and reverse strand reads (blue) are shown separately. y-axis: read count, x-axis: change in copy number relative to the reference plasmid. Reference copy numbers in each plasmid are in Additional file 1: Table S7. Black arrows: reads in these peaks may actually have shortened repeats

Fig. 3

a–j Distribution of predicted change in repeat copy number, for nanopore reads of human DNA with inserted repeats. Reads covering each of ten disease-associated repeat loci were selected, and the repeat region in each read was replaced by the repeat region of a plasmid nanopore read. y-axis: read count, x-axis: change in copy number relative to the reference human genome. Forward (red) and reverse strand reads (blue) are shown separately. Red arrows: projected repeat copy changes

Fig. 4

Distribution of predicted change in repeat copy number, for PacBio (RSII) reads of cloned SCA10 loci from 3 patients. a tandem-genotypes. Forward (red) and reverse strand reads (blue) are shown separately. b RepeatHMM with straightforward parameters. c RepeatHMM with non-obvious parameters suggested by its authors. y-axis: read count, x-axis: change in copy number relative to the reference human genome. Red arrow: expected repeat copy change, from McFarland et al. [9]

Fig. 5

PCR results and Sanger sequencing of two tandem-repeat loci in human genome NA12878, with expansions relative to the reference human genome (hg38). Reads were aligned by MUSCLE [33]. The histograms show read counts (y-axis) for predicted copy number changes (x-axis), with nanopore (rel3) in blue and PacBio (SRR3197748) in red. a Expansion of an inexact TATAT repeat in an intron of PCDH15: actually insertion of an AluYb8 SINE. b Expansion of an intergenic GT repeat: actually deletion in the reference human genome. The bands marked by asterisks were sequenced and proved to be non-target amplification

Fig. 6

Alignments of DNA reads (vertical) to the reference human genome (horizontal). Diagonal lines indicate alignments, of the same strands (red) and opposite strands (blue). The vertical stripes indicate repeat annotations in the reference genome: tandem repeats (purple) and transposable elements (pink). a Six reads from a BAFME patient that cover the disease-causing SAMD12 AAAAT repeat locus. b Close-ups of three reads with ~ 5 k expansions. c Two examples of chimeric human reads (rel3) with expanded CAG repeats at the ATXN7 disease locus

Fig. 7

a Prioritization of predicted repeat copy number changes in a BAFME patient. The BAFME expansion (AAATA: SAMD12 intron) is ranked 4th out of 0.7 million tandem repeats annotated in rmsk.txt. Forward (red) and reverse strand reads (blue) are shown separately. Histograms are raw output of tandem-genotypes-plot. b Prioritization of predicted repeat copy number changes in whole genome nanopore reads (NA12878 rel3) plus chimeric human/plasmid reads with pathological triplet-repeat expansions (AR, ATN1, ATXN2, ATXN3, ATXN7, CACNA1A, and HTT). These pathological expansions are prioritized within the top 10 out of 0.7 million tandem repeats in the genome, when compared to three control datasets using tandem-genotypes-join. c Comparison to control datasets with tandem-genotypes-join (joined) effectively de-prioritized other repeats, versus only using a single sample (single). y-axis: prioritization ranking

Fig. 8

a Genome-wide distribution of predicted change in repeat copy number, for nanopore MinION (rel3) and PacBio (SRR3197748) reads from the same human (NA12878). Nanopore tends to have negative and PacBio positive predicted changes, especially for short repeat units. Read number = 10,000 (randomly sampled). Nanopore reads are shown in blue and PacBio in red. b–d Genome-wide distribution of predicted change in repeat copy number for nanopore MinION (rel3) and PacBio (SRR3197748) reads from the same human (NA12878), and nanopore PromethION (ERR2585112-5) from a different individual (NA19240). b Distributions for AG di-nucleotide repeats. c Distributions for GAT. d Distributions for CTT. CTT shows the most prominent strand bias in nanopore rel3 reads among all types of triplet repeat (all types are in Additional file 1: Figures S9, S10, S13–S16). PromethION shows less strand bias compared to rel3. y-axis: read count, x-axis: change in copy number relative to the reference human genome