Household clustering and seasonal genetic variation of Plasmodium falciparum at the community-level in The Gambia

Reviewer #1 (Public review):

Summary:

The manuscript titled "Household clustering and seasonal genetic variation of Plasmodium falciparum at the community-level in The Gambia" presents a valuable genetic spatio-temporal analysis of malaria-infected individuals from four villages in The Gambia, covering the period between December 2014 and May 2017. The majority of samples were analyzed using a SNP barcode with the Spotmalaria panel, with a subset validated through WGS. Identity-by-descent (IBD) was calculated as a measure of genetic relatedness and spatio-temporal patterns of the proportion of highly related infections were investigated. Related clusters were detected at the household level, but only within a short time period.

Strengths:

This study offers a valuable dataset, particularly due to its longitudinal design and the inclusion of asymptomatic cases. The laboratory analysis using the Spotmalaria platform combined and supplemented with WGS is solid, and the authors show a linear correlation between the IBD values determined with both methods, although other studies have reported that at least 200 SNPs are required for IBD analysis. Data-analysis pipelines were created for (1) variant filtering for WGS and subsequent IBD analysis, and (2) creating a consensus barcode from the spot malaria panel and WGS data and subsequent SNP filtering and IBD analysis.

Weaknesses:

Further refining the data could enhance its impact on both the scientific community and malaria control efforts in The Gambia.

(1) The manuscript would benefit from improved clarity and better explanation of results to help readers follow more easily. Despite familiarity with genotyping, WGS, and IBD analysis, I found myself needing to reread sections. While the figures are generally clear and well-presented, the text could be more digestible. The aims and objectives need clearer articulation, especially regarding the rationale for using both SNP barcode and WGS (is it to validate the approach with the barcode, or is it to have less missing data?). In several analyses, the purpose is not immediately obvious and could be clarified.

(2) Some key results are only mentioned briefly in the text without corresponding figures or tables in the main manuscript, referring only to supplementary figures, which are usually meant for additional detail, but not main results. For example, data on drug resistance markers should be included in a table or figure in the main manuscript.

(3) The study uses samples from 2 different studies. While these are conducted in the same villages, their study design is not the same, which should be addressed in the interpretation and discussion of the results. Between Dec 2014 and Sept 2016, sampling was conducted only in 2 villages and at less frequent intervals than between Oct 2016 to May 2017. The authors should assess how this might have impacted their temporal analysis and conclusions drawn. In addition, it should be clarified why and for exactly in which analysis the samples from Dec 2016 - May 2017 were excluded as this is a large proportion of your samples.

(4) Based on which criteria were samples selected for WGS? Did the spatiotemporal spread of the WGS samples match the rest of the genotyped samples? I.e. were random samples selected from all times and places, or was it samples from specific times/places selected for WGS?

(5) The manuscript would benefit from additional detail in the methods section.

(6) Since the authors only do the genotype replacement and build consensus barcode for 199 samples, there is a bias between the samples with consensus barcode and those with only the genotyping barcode. How did this impact the analysis?

(7) The linear correlation between IBD-values of barcode vs genome is clear. However, since you do not use absolute values of IBD, but a classification of related (>=0.5 IBD) vs. unrelated (<0.5), it would be good to assess the agreement of this classification between the 2 barcodes. In Figure S6 there seem to be quite some samples that would be classified as unrelated by the consensus barcode, while they have IBD>0.5 in the Genome-IBD; in other words, the barcode seems to be underestimating relatedness.
a. How sensitive is this correlation to the nr of SNPs in the barcode?

(8) With the sole focus on IBD, a measure of genetic relatedness, some of the conclusions from the results are speculative.
a. Why not include other measures such as genetic diversity, which relates to allele frequency analysis at the population level (using, for example, nucleotide diversity)? IBD and the proportion of highly related pairs are not a measure of genetic diversity. Please revise the manuscript and figures accordingly.
b. Additionally, define what you mean by "recombinatorial genetic diversity" and explain how it relates to IBD and individual-level relatedness.
c. Recombination is one potential factor contributing to the loss of relatedness over time. There are several other factors that could contribute, such as mobility/gene flow, or study-specific limitations such as low numbers of samples in the low transmission season and many months apart from the high transmission samples.
d. By including other measures such as linkage disequilibrium you could further support the statements related to recombination driving the loss of relatedness.

(9) While the authors conclude there is no seasonal pattern in the drug-resistant markers, one can observe a big fluctuation in the dhps haplotypes, which go down from 75% to 20% and then up and down again later. The authors should investigate this in more detail, as dhps is related to SP resistance, which could be important for seasonal malaria chemoprofylaxis, especially since the mutations in dhfr seem near-fixed in the population, indicating high levels of SP resistance at some of the time points.

(10) I recommend that raw data from genotyping and WGS should be deposited in a public repository.

Reviewer #2 (Public review):

Summary:

Malaria transmission in the Gambia is highly seasonal, whereby periods of intense transmission at the beginning of the rainy season are interspersed by long periods of low to no transmission. This raises several questions about how this transmission pattern impacts the spatiotemporal distribution of circulating parasite strains. Knowledge of these dynamics may allow the identification of key units for targeted control strategies, the evaluation of the effect of selection/drift on parasite phenotypes (e.g., the emergence or loss of drug resistance genotypes), and analyze, through the parasites' genetic nature, the duration of chronic infections persisting during the dry season. Using a combination of barcodes and whole genome analysis, the authors try to answer these questions by making clever use of the different recombination rates, as measured through the proportion of genomes with identity-by-descent (IBD), to investigate the spatiotemporal relatedness of parasite strains at different spatial (i.e., individual, household, village, and region) and temporal (i.e., high, low, and the corresponding the transitions) levels. The authors show that a large fraction of infections are polygenomic and stable over time, resulting in high recombinational diversity (Figure 2). Since the number of recombination events is expected to increase with time or with the number of mosquito bites, IBD allows them to investigate the connectivity between spatial levels and to measure the fraction of effective recombinational events over time. The authors demonstrate the epidemiological connectivity between villages by showing the presence of related genotypes, a higher probability of finding similar genotypes within the same household, and how parasite-relatedness gradually disappears over time (Figure 3). Moreover, they show that transmission intensity increases during the transition from dry to wet seasons (Figure 4). If there is no drug selection during the dry season and if resistance incurs a fitness cost it is possible that alleles associated with drug resistance may change in frequency. The authors looked at the frequencies of six drug-resistance haplotypes (aat1, crt, dhfr, dhps, kelch13, and mdr1), and found no evidence of changes in allele frequencies associated with seasonality. They also find chronic infections lasting from one month to one and a half years with no dependence on age or gender.

The use of genomic information and IBD analytic tools provides the Control Program with important metrics for malaria control policies, for example, identifying target populations for malaria control and evaluation of malaria control programs.

Strength:

The authors use a combination of high-quality barcodes (425 barcodes representing 101 bi-allelic SNPs) and 199 high-quality genome sequences to infer the fraction of the genome with shared Identity by Descent (IBD) (i.e. a metric of recombination rate) over several time points covering two years. The barcode and whole genome sequence combination allows full use of a large dataset, and to confidently infer the relatedness of parasite isolates at various spatiotemporal scales.

Reviewer #3 (Public review):

This study aimed to investigate the impact of seasonality on the malaria parasite population genetic. To achieve this, the researchers conducted a longitudinal study in a region characterized by seasonal malaria transmission. Over a 2.5-year period, blood samples were collected from 1,516 participants residing in four villages in the Upper River Region of The Gambia and tested the samples for malaria parasite positivity. The parasites from the positive samples were genotyped using a genetic barcode and/or whole genome sequencing, followed by a genetic relatedness analysis.

The study identified three key findings:

(1) The parasite population continuously recombines, with no single genotype dominating, in contrast to viral populations;

(2) The relatedness of parasites is influenced by both spatial and temporal distances; and

(3) The lowest genetic relatedness among parasites occurs during the transition from low to high transmission seasons. The authors suggest that this latter finding reflects the increased recombination associated with sexual reproduction in mosquitoes.

The results section is well-structured, and the figures are clear and self-explanatory. The methods are adequately described, providing a solid foundation for the findings. While there are no unexpected results, it is reassuring to see the anticipated outcomes supported by actual data. The conclusions are generally well-supported; however, the discussion on the burden of asymptomatic infections falls outside the scope of the data, as no specific analysis was conducted on this aspect and was not stated as part of the aims of the study. Nonetheless, the recommendation to target asymptomatic infections is logical and relevant.