This paper addresses the protein classification problem, and explores how its accuracy can be improved by using information from time-course gene expression data. The methods are tested on data from the most deadly species of the parasite responsible for malaria infections, Plasmodium falciparum. Even though a vaccination for Malaria infections has been under intense study for many years, more than half of Plasmodium proteins still remain uncharacterized and therefore are exempted from clinical trials. The task is further complicated by a rapid life cycle of the parasite, thus making precise targeting of the appropriate proteins for vaccination a technical challenge. We propose to integrate protein-protein interactions (PPIs), sequence similarity, metabolic pathway, and gene expression, to produce a suitable set of predicted protein functions for P. falciparum. Further, we treat gene expression data with respect to various changes that occur during the five phases of the intraerythrocytic developmental cycle (IDC) (as determined by our segmentation algorithm) of P. falciparum and show that this analysis yields a significantly improved protein function prediction, e.g., when compared to analysis based on Pearson correlation coefficients seen in the data. The algorithm is able to assign "meaningful" functions to 628 out of 1439 previously unannotated proteins, which are first-choice candidates for experimental vaccine research.