We compute (algebraically) the Euler characteristic of a complex of sheaves with constructible cohomology. A stratified Poincaré-Hopf formula is then a consequence of the smooth Poincaré-Hopf theorem and of additivity of the Euler-Poincaré characteristic with compact supports, once we have a suitable definition of index.

SM-11044 is the only β-adrenergic agonist that inhibits guinea pig eosinophil chemotaxis and induces relaxation of depolarized rat colon tonus. We have previously reported the purification of a 34 kDa photoaffinity-labeled SM-11044 binding protein (SMBP) from rat colon that may mediate the biological effects of the ligand and that differs from all known monoamine receptors (Sugasawa et al., J. Biol. Chem. 272 (1997) 21244). The present report describes partial amino acid sequence of rat SMBP and molecular cloning of corresponding human SMBP (hSMBP) cDNA. This cDNA encodes a 588 amino acid residue polypeptide comprising a signal peptide, a long hydrophilic amino-terminal region, and a highly hydrophobic C-terminal portion organized into nine putative transmembrane domains. The sequence and structure of hSMBP shows homology to members of a new transmembrane protein 9 superfamily (TM9SF). Comparison of hSMBP with related protein sequences from yeast, plant and human revealed two subgroups within TM9SF. The members of these groups differ in length and have characteristic amino acid sequence motifs in their amino-terminal portion. Northern blot analysis revealed two major SMBP mRNAs, at 3.4 and 3.8 kb, that were present in all the human tissues examined. Western blot experiments detected SMBP as a 70 kDa protein that may be further cleaved into an active 34 kDa N-terminal polypeptide. Stable Chinese Hamster Ovary cell transfectants expressing hSMBP cDNA displayed specific binding of [125I]iodocyanopindolol that was displaced by SM-11044 in a dose-dependent manner. Thus, SMBP is the first member of TM9SF with functional ligand binding properties, suggesting that some of these integral membrane proteins may function as channels, small molecule transporters or receptors.

Benchmarking and performance evaluation of high performance computing are a continuously on-going process that consumes significant staff time and generates large amounts of data to be stored and analyzed. The goal of this work is to propose an automated method to generate, capture and analyze an extensive performance profile. Although our initial efforts focus on commodity clusters, they are applicable to any parallel or distributed high performance computing system. Such an automated system has three main components: (1) a data collection and storage module, (2) a data analysis module, and (3) a run-time execution module. The focus of this report is on the data collection and storage module.