SCOTTI: Inferring transmission with the Structured Coalescent - PowerPoint PPT Presentation

SCOTTI: Inferring transmission with the Structured Coalescent Nicola De Maio, Chieh-Hsi Wu, Daniel Wilson

Host information ACGTCG Time ACATCG ACGTTG CCGTTG

Complications: within-host coalescent Transmission and evolution history Phylogenetic Transmission network tree

Complications Within-host coalescent Incomplete bottleneck B H2 H3 A H1 H3 Time Time S3 H2 H1 S2 S1 S3 S1 S2 H1 H2 H3 H3 H1 H2 S1 S1 S2 S3 S2 S3 D C Multiple infections Non-sampled case H1 H1 H3 H2 H3 S1 Time Time S1 H2 S3 S2 S3 H1 H2 H3 H1 H3 H2 S1 S3 S3 S1 S2

Structured coalescent Coalescence events only within demes Migration moves single lineages between demes. We use a recent efficient approximation to the structured coalescent: BASTA (De Maio et al 2015 PLOS Genetics).

Phylogeography ¡with ¡BASTA ¡ (a)(Mugra7on( (b)(Mul7TypeTree( (c)(BaStA( Posterior(median(and(95%(interval( 1e+06 1e+06 1e+06 1e+03 1e+03 1e+03 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● 1e+00 1e+00 ● ● ● ● ● ● ● 1e+00 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 1e − 03 1e − 03 1e − 03 ● ● ● (((Calibra7on:(95%( (((Calibra7on:(56%( (((Calibra7on:(87%( ((Correla7on:(0.58( ((Correla7on:(0.77( ((Correla7on:(0.83( 1e − 06 1e − 06 1e − 06 1e − 03 1e − 01 1e+01 1e+03 1e − 03 1e − 01 1e+01 1e+03 1e − 03 1e − 01 1e+01 1e+03 Migra7on(rates(ra7o( Migra7on(rates(ra7o( Migra7on(rates(ra7o( Discrete trait Structured coalescent BASTA (Lemey et al 2009) (Vaughan et al 2014) (De Maio et al 2015) fast but inaccurate. accurate but slow. accurate and fast.

SCOTTI SCOTTI: Efficient Reconstruction of Transmission within Outbreaks with the Structured Coalescent Nicola De Maio 1,2 � , Chieh-Hsi Wu 2 , Daniel J Wilson 1,2,3 - Hosts (demes) have same population size. � - Hosts have limited lifespan. ���� ������ - No bottlenecks at transmission. - Lineages do not migrate together at transmission. ���� ������

Simulated model Extension of: Didelot et al 2014 MBE. Hall et al 2015 PLOS Comput Biol. - Bottlenecks at transmission. - Lineages migrate together at transmission. - Only one transmission per host.

Benchmark model Jombart et al 2014 PLOS Comput Biol - No within-host population. - No within-host evolution. - Mutations accumulate at transmission. - Only one sample per host. - Generations of same lengths.

Models used Multispecies SCOTTI Outbreaker (simulated history) (ideal inference) (ideal inference) A B H3 C H1 H1 H1 H3 S1.1 H2 H2 H2 H3 H4 H4 S1.1 S1.1 Time S2.1 S1.2 S1.2 H4 S4.1 S4.1 S2.1 S4.1 S2.1 S4.2 S4.2 S2.2 Jombart et al 2014 S2.2 PLOS Comput Biol Didelot et al 2014 MBE De Maio et al 2016 PLOS Comput Biol Hall et al 2015 PLOS Comput Biol

Calibration 95% posterior sets Accuracy of point estimate Simulations Outbreaker

Calibration 95% posterior sets Accuracy of point estimate Simulations Outbreaker

Mean*accuracy** A* ula'on*scenario* of*the*point*es'mate* Calibra'on* Within-host similarity Simula' B* A* Simulations Calibra'on*

Simulations – Running time Genera4ons" 3" 5" 7" 1"sample"" 1 6000" 6000" 6000" 6000" "per"host" 4000" 4000" 4000" 4000" 2"samples"" 3" 2 Hosts"per"genera4on" ""per"host" 118"s" 239"s" 171"s" 346"s" 2000" 2000" 2000" 2000" 100"s" 124"s" 0" 0" 0" 0" 6000" 6000" 6000" 4000" 4000" 4000" 1733"s" 547"s" 5" 314"s" 774"s" 233"s" 385"s" 2000" 2000" 2000" 0" 0" 0" 6000" 6000" 6000" 4039"s" 1279"s" 2021"s" 4000" 4000" 4000" 822"s" 7" 414"s" 819"s" 2000" 2000" 2000" 0" 0" 0"

FMDV data 0.13 IP8 5/1 IP6b IP8 0.00 IP6b 0.00 0.11 SCOTTI 9 IP7 3 . 0 0.15 2/1 0 3 0.16 . 3/1 0 0 0 . . 5 3 Outbreaker 4 7 IP7 0.00 IP3b 0.43 2 3 0 . 3/1 IP4b IP4b 0.00 0 . 4 2 0 . 1 IP3b 0.00 9 2 6 0 . 10/1 0 . 2 5 0 . 1 5 IP3c 0.00 1/1 IP3c 0.20 0.21 6 1 . 0 0.16 0.16 IP1b 0.11 IP5 2/1 6/1 IP2b IP5 0.01 0.37 2/1 IP2b 0.25 IP2c 0.23 0 . 3 2 1 2 0 . 0 . 4 6 4 1 . 0 0.17 IP1b 0.53 Cottam et al. 2008 PLOS Pathogens 0.62 IP2c 0.15

K. Pneumoniae data SCOTTI 25 0.98 PMK1 PMK5 H30 2 0.23 1 . 0 21 0.52 26 22 0.98 24 3 PMK3 1 PMK7 . 0 1 0.24 0 . 0 6 0.48 PMK6 0.98 PMK9 PMK4 23 H30 0.98 18 19 PMK26 0.11 8 2 PMK10 0 . 1 PMK23 0.77 1 PMK13 PMK15 Outbreaker 7 0.20 20 0 0.06 . 0.07 0 16 0.27 0.23 1 0.31 5 2 0 . PMK18 14 PMK12 4 2 . 0 PMK17 PMK21 0.55 0.96 0.30 1 1 PMK11 15 17 0.41 PMK16 PMK25 0.33 0.5 0.66 0.06 6 0.51 0.5 0.46 12 3 0.49 0.5 0.5 1 PMK20 PMK14 PMK19 4 0.45 0.5 PMK24 0.5 11 PMK22 5 0.26 0.25 0.49 0.98 7 0.33 1 0.35 0.99 10 9 13 Stoesser et al. 2014 Antimicrobial agents and chemotherapy

K. Pneumoniae data PMK1 [0,2]& PMK3 PMK4 [0,2]& SCOTTI PMK4 NS PMK4 [1,2]& PMK4 PMK7 NS NS PMK5 NS PMK5 [0,9]& PMK5 PMK6 [1,2]& [0,2]& PMK9 [0,2]& PMK10 NS [1,2]& PMK11 NS [0,2]& PMK11 NS [0,2]& [0,2]& PMK12 NS PMK13 PMK13 [0,2]& PMK13 [0,2]& PMK14 [1,2]& PMK16 [0,2]& PMK17 [1,2]& NS NS [0]& PMK18 NS PMK21 NS PMK21 NS PMK21 NS [0,2]& PMK21 PMK21 PMK21 PMK21 PMK21 PMK22 PMK21 PMK22 [0,4]& PMK24 NS [1,2]& PMK21 [0]& NS PMK25 [1,2]& PMK21 PMK19 [0,2]& NS [0,2]& PMK23 [0,2]& NS PMK26 NS PMK26 PMK26 [0,2]& [1,2]& PMK20 [1,2]& [1,13]& PMK15 H30

Summary SCOTTI Different models result in different inferences. New inference of transmission in BEAST2: SCOTTI. Future work: transmission bottlenecks, introductions, epidemiological models.

Thanks for listening! Daniel J Wilson Chieh-Hsi Wu Crook group (NDM Microbiology)