DBLα-tags sequence groupings and common names used in the text.

Sample inclusion and syndrome classification.

A). Flowchart of children whose samples were included in the study and analysis. B). Venn diagram of severe malaria syndromes among admissions included in the study. IC: impaired consciousness, RD: Respiratory distress, SMA: severe malarial anaemia.

Associations between var gene expression and severe malaria syndromes.

A-C Results from five independent regression models in which cys2 or its subgroups (group A-like, cp1, cp2, and cp3) were considered as single dependent variables with the severe malaria syndromes, impaired consciousness, respiratory distress and severe malaria anaemia as explanatory variables, while adjusting for age and parasitemia. D-F Results from twelve independent regression models considering the transcript levels for gpA1-gpC as sole dependent variables, with impaired consciousness, respiratory distress and severe malaria anaemia as explanatory variables while adjusting for age and parasitaemia. Regression coefficient (β) and corresponding 95% confidence intervals (CI) were plotted.

Relationship between VEH and parasite or host factors

A). Distribution of Variant Expression Homogeneity (VEH) values across pure clinical malaria groups in children. The y-axis shows VEH scores (higher values show more homogenous var expression); x-axis show clinical presentation. B). Comparison of VEH between SMA and non-SMA cases. C). Spearman’s correlation between VEH and parasite factors, coefficient >0 indicate a positive association while coefficient <0 shows a negative association. D). Levels of antibodies against infected erythrocytes across clinical malaria presentation. E). Caregiver-reported fever duration (days) at admission. Horizontal brackets indicate significant pairwise differences (Wilcoxon rank-sum test).* <0.05, ** <0.001, *** pvalue <0.0001, **** pvalue <0.0001. α-IE; antibodies to Infected erythrocytes, ic: Impaired consciousness, rd: Respiratory distress, sma: Severe malarial anaemia.

Relationship between rosetting, group A var gene expression and malaria

A). Mean percentage rosetting frequency between severe and non-severe malaria. NS: non-severe malaria, SM: Severe malaria. B). Association between rosetting and severe malaria syndromes. C) Spearman correlation coefficients (with 95% confidence intervals) between the frequency of rosetting and expression levels of different var gene subgroups. Red points indicate statistically significant correlations (p < 0.05). Subgroups include (cys2, groupA-like, cp1–cp3) from DBLα-tag data and gpA1, gpA2, dc8-1,-dc84, dc13, dc9, gpB and gpC from RT-qPCR data. D and E) Logistic regression models assessing the association between rosetting and malaria syndromes after adjusting for group A var genes. D) Respiratory distress, and E) severe malarial anaemia. The blue dotted line represents odds ratio for rosetting on a specific syndrome before adjusting for var gene expression. Red error bars indicate statistically significant associations (p < 0.05).

Relationship between group A var genes and rosetting in different malaria syndromes using logistic regression models.

Models 1-4 are unadjusted while models 5 to 7 are adjusted for either rosetting or var gene expression levels. Bold pvalues are significant at alpha less than 0.05

group A-like expression levels in relation to death

A) Group A–like expression in children who died (n = 40) versus those who survived (n = 614). P is a pvalue obtained from logistic regression model. B) Expression of Group A–like expression across clinical malaria syndromes; Red points indicate children who died. Number of deaths by syndrome: mild=0, moderate = 5, IC = 20, SMA = 1, overlap = 14, RD = 0. IC: impaired consciousness, RD: respiratory distress, SMA: severe malarial anaemia, overlap: mixed malaria infection with two or all the three severe malaria syndromes.

Total malaria admissions and mean patient age per year.

Number of patients per year (left axis), mean patient age per year (right axis). The solid black line is the average patient age of total admission, the dashed line is the average patient age in this study

Associations between var gene expression and pure severe malaria syndromes.

A-C Results from five regression models in which cys2 or its subgroups (group A-like, cp1, cp2, and cp3) were considered as single dependent variables with the pure severe malaria syndromes, impaired consciousness, respiratory distress and severe malaria anaemia as explanatory variables, while adjusting for age and parasitemia. D-E Results from twelve regression models considering the transcript levels for gpA1-gpC as sole dependent variables, with and impaired consciousness, respiratory distress and severe malaria anaemia as explanatory variables while adjusting for age and parasitemia. Regression coefficient (β) and corresponding 95% confidence intervals (CI) were plotted.

Description of Primers used in quantifying var genes.

Clinical Severity Comparison vs Non-Severe

Associations between var gene expression and severe malaria syndromes from linear model.

Each regression model (1-17) included; Impaired consciousness, Respiratory distress and severe malaria anaemia as explanatory variables while adjusting for age and parasitemia. This model allowed for adjusting for overlapping syndromes. *shows significant adjusted p-value

Associations between var gene expression and non-overlapping severe malaria syndromes from linear model.

Each regression model (1-17) included; Impaired consciousness, Respiratory distress and severe malaria anaemia as explanatory variables while adjusting for age and parasitemia. *shows significant adjusted p-value

Expression of pfsir2a among clinical syndromes.

Pvalues are derived from Wilcoxon rank-sum test

The median body temperature within each syndrome at admission

Association of var gene expression, and mortality.

p-value<0.05 are highlighted