Malaria - screening hits to kill the parasite

The development of drugs to treat malaria has been slow, too slow in fact to keep pace with the quickly adapting parasite that has developed mechanisms of resistance against all current treatments.
Nature now features two articles that release screening results against Plasmodium falciparum, the malaria causing blood parasite. One comes from GlaxoSmithKline, the other from an academic and clinical consortium (Guy et al).
Guy et al have tested ~300.000 compounds and GSK have screened what I believe is their entire library of just about 2mio compounds against the laboratory malarial strain 3D7. Both assay systems are cell based and hence do not associate compounds with explicit protein targets.
GSK identifies 13,533 compounds as confirmed hits (with data deposited in ChEMBL) and Guy et al report dose-response curves for 1,134 validated hits. Guy et al go on cherry-picking 172 of their reported hits for cross validation against 66 Plasmodium proteins including the hemzoin-forming pathways, dehydroorotate dehydrogenase and falcipain. They analyze all results based on a tree that is constructed from compound similarities and looks very funky.
GSK keep it cheap but systematic and hypothesize about molecular targets based on a scheme that clusters compounds according to similarity and associates these clusters with targets that have been reported in previous assays for any compound in the cluster. Most of these targets are in fact human proteins and therefore the Plasmodium targets are inferred by sequence homology. The hypothesis that similar compounds bind the same target seems a bit naive but to be fair is likely to be the most practical approach to handle the very large number of reported hits.

One might get carried away comparing the different approaches of a pharmaceutical company and an academic consortium and project a few stereotypes but let's not do that! Both papers tell a story with open end, failing to comprehensively explore (and exploit) the set of reported hits. Cheer up, resolve is that they provide many opportunities for drug discovery!
Clearly there are limitations to the exploitability of this data as most of the reported interactions are on a cellular level with only a minority of compounds passed on to protein-level screens. Yet both papers rock as finally efforts are made to tackle malaria with structures deposited in open access databases.