HIV-1 drug resistance testing using Oxford Nanopore’s MinION – a long-range low error sequencing approach Alexander Thielen AREVIR 2017
Oxford Nanopore’s technology
Oxford Nanopore’s technology – long range sequencing
Oxford Nanopore’s technology – long range sequencing – very fast
Oxford Nanopore’s technology
Oxford Nanopore’s technology – long range sequencing – very fast – portability / global access (search for “Massimo Delledonne: MinIONs and Nanofrogs“)
Oxford Nanopore’s MinION
Oxford Nanopore’s technology – long range sequencing – very fast – portability / global access (search for “Massimo Delledonne: MinIONs and Nanofrogs“)
Oxford Nanopore’s technology – long range sequencing – very fast – portability / global access (search for “Massimo Delledonne: MinIONs and Nanofrogs“) – relatively cheap, no capital cost – but: high error rate
Oxford Nanopore’s error rate
Error rate reduction using an Rolling Circle Amplification approach
RCA Approach: 1. Amplification MID1 M13for 3.1kb cDNA PR RT RNaseH INT M13rev MID2 PCR product
RCA Approach: 2. Circularization Self-ligation using T4 ligase, digestion of non-circularized fragments
RCA Approach: 3. Rolling Circle Amplification random priming isothermal amplification by phi29 polymerase
RCA Approach: 3. Rolling Circle Amplification Final products: copy 1 copy 2 copy n 80-100 kb Amplification products are sheared to ~20kb
RCA Approach: 4. Library Preparation & Sequencing
RCA Approach: 5. Analysis copy 1 copy 2 copy n 80-100 kb Individual copies are detected in the sequenced reads and aligned against each other x x copy 1 x x x copy 2 ... x x x copy n A consensus sequence is generated consensus
RCA accuracy estimation – except for homopolymers, errors shoud occur randomly – consensus sequences should therefore reduce error rates – accuracy simulations: Amplicon R7 sim. Estimated RCA Accuracy count 1 75.00% 80.00% 85.00% 90.00% 77.91% 3 84.38% 89.60% 93.93% 97.20% 84.41% 5 89.65% 94.21% 97.34% 99.14% 89.38% 7 92.94% 96.67% 98.79% 99.73% 91.49% 9 95.11% 98.04% 99.44% 99.91% 93.03%
Experiment – 16 samples were prepared with the RCA approach (RCA- 1D) – for comparison, the same amplified 3.1kb products were also sequenced with • Illumina MiSeq, with Nextera XT tagmentation (MiSeq) • MinION with 1D Ligation Kit without fragmentation & RCA (1D- only) – MiSeq results were used as gold-standard
Results – MinION output: • 1D-only: 240.000 reads mapped onto HIV-1 1300 bp (range: 31-3098 bp) • RCA-1D: 198.000 reads mapped onto HIV-1 1698 bp (range: 173-82.556 bp) 3000 Frequency 2000 1000 0 1 2 3 4 5 log10 read length
Results – Error rates (when only one copy of the amplified region is considered): • 1D-only: 14.7% in total, 6.4% substitution errors, 5.6% deletions, and 1.1% insertions • RCA-1D: 14.7% in total, 6.0% mismatches, 6.3%, deletions 0.8% insertions RCA has no effect on raw error rates
Results – median RCA consensus error rates: region copies total substitution deletion insertion 1 14.72% 6.03% 6.29% 0.77% 2 11.25% 5.63% 3.18% 1.00% 3 11.00% 3.49% 6.02% 0.18% 4 7.39% 1.64% 4.76% 0.00% 5 6.12% 1.22% 3.97% 0.00% 6 4.44% 0.94% 2.79% 0.00% 7 2.76% 0.00% 1.56% 0.00% 8 1.78% 0.00% 0.00% 0.00% 9 1.56% 0.00% 0.00% 0.00% 10 1.47% 0.00% 0.00% 0.00% 11 1.2% 0.00% 0.00% 0.00%
Results error rates 16,00% 14,00% 12,00% 10,00% 8,00% 6,00% 4,00% 2,00% 0,00% 0 1 2 3 4 5 6 7 8 9 10 11 12 -2,00% total substitution deletion insertions
RCA accuracy estimation vs. experiment Amplicon Estimated RCA Accuracy R7 sim. R9 exp. count 1 75.00% 80.00% 85.00% 90.00% 77.91% 85.28% 3 84.38% 89.60% 93.93% 97.20% 84.41% 89.00% 5 89.65% 94.21% 97.34% 99.14% 89.38% 93.88% 7 92.94% 96.67% 98.79% 99.73% 91.49% 97.24% 9 95.11% 98.04% 99.44% 99.91% 93.03% 98.44%
Results – Basecalling errors at different sensitivity cut-offs: 35,00% 30,00% 25,00% 20,00% 15,00% 10,00% 5,00% 0,00% 0 2 4 6 8 10 12 5% 10% 15% 20% – Basecalling errors almost entirely within or adjacent to homo-polymers
Outlook: Long-range covariation analysis 16-76608: M184V, K219R, E138K ~ 100% V108I: 77% V179I: 30% H221Y: 24% V106I: 4.4%
Acknowledgments Kirsten Becker Nina Engel Anja Förster Anna Memmer Bettina Spielberger Martin Däumer Bernhard Thiele
Recommend
More recommend