Enhancing Gene Co-Expression Network Inference for the Malaria Parasite Plasmodium falciparum.

Area of Science:

• Genomics
• Bioinformatics
• Parasitology

Background:

• Malaria causes over 550,000 deaths annually, primarily due to drug resistance in *Plasmodium falciparum*.
• Despite a published genome, 44.6% of *P. falciparum* genes have unknown functions, hindering drug target identification and resistance studies.

Purpose of the Study:

• To improve functional annotations of *P. falciparum* genes by analyzing gene co-expression networks.
• To systematically predict functional annotations for all genes in *P. falciparum* using multiple inference methods.
• To identify potential drug targets and understand drug resistance evolution.

Main Methods:

• Construction and evaluation of multiple gene co-expression networks for *P. falciparum*.
• Utilizing network clustering and leave-one-out cross-validation to assess prediction accuracy against known Gene Ontology (GO) terms.
• Analyzing the complementarity and overlap of network edges and predicted functional knowledge across different inference methods.

Main Results:

• Gene co-expression networks demonstrated high precision (up to 87%) in predicting functional annotations, but low recall (below 15%).
• Network edges were largely complementary, with 47-85% unique to each inferred network.
• Annotation predictions were also complementary, with a maximum pairwise overlap of only 27%.

Conclusions:

• Different network inference methods capture distinct aspects of *P. falciparum* biology.
• A single network inference method is insufficient for comprehensive functional annotation; combining methods is recommended.
• The study provides valuable ranked lists of gene interactions and predicted annotations for the malaria research community.