Development of amplicon deep sequencing markers and data analysis pipeline for genotyping multi-clonal malaria infections

Abstract

Background: Amplicon deep sequencing permits sensitive detection of minority clones and improves discriminatory power for genotyping multi-clone Plasmodium falciparum infections. New amplicon sequencing and data analysis protocols are needed for genotyping in epidemiological studies and drug efficacy trials of P. falciparum.

Methods: Targeted sequencing of molecular marker csp and novel marker cpmp was conducted in duplicate on mixtures of parasite culture strains and 37 field samples. A protocol allowing to multiplex up to 384 samples in a single sequencing run was applied. Software "HaplotypR" was developed for data analysis.

Results: Cpmp was highly diverse (H_e = 0.96) in contrast to csp (H_e = 0.57). Minority clones were robustly detected if their frequency was >1%. False haplotype calls owing to sequencing errors were observed below that threshold.

Conclusions: To reliably detect haplotypes at very low frequencies, experiments are best performed in duplicate and should aim for coverage of >10'000 reads/amplicon. When compared to length polymorphic marker msp2, highly multiplexed amplicon sequencing displayed greater sensitivity in detecting minority clones.

Keywords: Amplicon sequencing; HaplotypR software; Haplotype clustering; Malaria; Multi-clone infections; Plasmodium falciparum; SNP; cpmp; csp; msp2.

Fig. 1

Mismatch rate per nucleotide position derived from all samples sequenced for markers cpmp and csp. Each data point represents the mean observed mismatch rate observed in all reads of one sample at the respective nucleotide position. Red data points: control samples (P. falciparum culture strains); black data points: field samples; X-axis: nucleotide position in sequenced fragment; Y-axis: mismatch rate with respect to the reference sequence (for control samples: sequences of strains 3D7 and HB3, for field samples, 3D7 sequence); dashed grey lines represent SNPs with a mismatch rate of >0.5 in >1 sample; red dotted horizontal line indicates a mismatch rate of 0.5; solid black vertical line: position of concatenation of forward and reverse reads

Fig. 2

Simulation of minority clone detectability by bootstrapping for marker cpmp and csp. Cut-off for acceptance of a haplotype was a minimum coverage per haplotype of 3 reads and a minority clone detection limit of 1:1000. Samples were drawn from reads of defined mixtures of P. falciparum strain 3D7 and HB3. X-axis shows dilution ratios of strains 3D7 and HB3; Y-axis indicates the sampling size (number of draws from sequence reads) for each mixture of strains. Sampling was repeated 1000 times to estimate mean minority clone detectability

Fig. 3

Comparison of genotyping by length-polymorphic marker msp2 and amplicon sequencing of cpmp and csp. Raw data from length-polymorphism- and SNP-based genotyping for one P. falciparum-positive field sample. Left panel: Capillary electropherograms (CE) for msp2 nested PCR products (duplicate experiments); X-axis: fragment length, Y-axis: peak heights (arbitrary intensity units); size standards: red/orange peaks; 3D7-type msp2 genotypes: green peaks; FC27-type msp2 genotypes: blue peaks. Middle and right panel: Dendrograms derived from sequence reads of marker cpmp (middle) and csp (right); coloured lines represent membership to a specific, colour-coded haplotype; Grey lines: sequence reads of PCR artefacts (later excluded by cut-off settings); line length: number of mismatches according to bar insert. Bottom panels: Read counts (n) and percentage of reads (%) per haplotype and final multiplicity call

Fig. 4

Frequency distribution of multiplicity of infection and allelic frequencies of cpmp, csp and msp2-CE. 37 P. falciparum positive samples from PNG were analysed for the 3 markers cpmp (27 haplotypes), csp (4 haplotypes) and msp2-CE (25 genotypes). Pie charts represent allelic frequency distribution for each marker in 37 samples

Fig. 5

Bioinformatic analysis pipeline applied on highly multiplexed deep sequencing data