Table 2. Comparison of hits between the three studies.

These are results from the R package SuperExactTest, which statistically tests if there is an enrichment of overlapping genes in multiple sets ²⁵. The expected overlap is the null hypothesis that there is only random sampling of genes. The p-value is the one-tail probability of observing equal or larger overlapping genes at random. The background size was set to 5,800, which is how many transcripts there are in v29 of the 3D7 transcriptome. For comparison, we used hits from Brancucci et al. (bulk) ¹⁶, this study (scRNA-seq data in Supplementary File 4, all hits with positive fold change in last two time points; sc) and Poran et al. ¹⁵ (scRNA-seq data in table S3, all hits; Poran).

Intersections	Degree	Observed overlap	Expected overlap	FE	p-value
Brancucci et al. 16 (bulk; Supplementary File 2, TP3/4)	1	268
Poran et al. 15 (Table S3)	1	28
Brancucci et al. (sc; Supplementary File 4 + FC)	1	125
Poran & bulk	2	9	1.29	6.96	2.67E-06
Poran & sc	2	3	0.60	4.97	0.022
sc & bulk	2	14	5.78	2.42	0.0017
Poran & sc & bulk	3	2	0.03	71.73	3.64E-04

Fe, fold enrichment over null; FC, fold change; sc, single cell.