Multicohort analysis identifies conserved transcriptional interactions between humans and Plasmodium falciparum

<h2>Abstract</h2> Malaria is caused by <i>Plasmodium</i>, a parasite that replicates inside and ruptures erythrocytes, causing an intense inflammatory response. Advances in high-throughput sequencing technologies have enabled the simultaneous study of the gene expression in humans and <i>P. falciparum</i>. However, the high-dimensional correlational networks generated in previous studies challenge the interpretation of the underlying biology, whereas associations found in one cohort might not replicate in independent samples due confounding factors affecting gene expression. We combined multicohort analysis of correlations with a hierarchical grouping approach to improve the discovery and interpretation of transcriptional associations between humans and <i>P. falciparum</i>. We analyzed nine public dual-transcriptomes acquired from whole blood of individuals infected with <i>P. falciparum</i>. Blood Transcription Modules (BTM) were used to reduce the dimension of host transcriptomes and Spearman's correlation analysis was used to identify host-parasite associations. Following, we performed meta-analysis of correlations with Stouffer's method and Bonferroni correction that resulted in a major transcriptional meta-network between humans and <i>P. falciparum</i>. We identified, for example, positive correlations between <i>PAK1, NFKBIA, BIRC2, NLRC4, TLR4, RIPK2</i> expression and <i>PF3D7_1205800</i>, a putative <i>P. falciparum</i> high mobility group protein B3 (HMGB3). We also applied a leave-one-out strategy to prevent influence of confounding factors, resulting in highly conserved associations between host genes related to inflammation, immune cells, and glycerophospholipid metabolism with <i>PF3D7_1223400</i>, which encodes a putative phospholipid-transporting ATPase. Paired metabolomics and transcriptomics data revealed negative correlation between <i>PF3D7_1223400</i> expression and the relative abundance of 1-linoleoyl-GPG. Collectively, our study provides data-driven hypotheses about molecular mechanisms of host-parasite interaction.