Publications

The structure, function, stability, and many other properties of a protein in a fixed environment are fully specified by its sequence, but in a manner that is difficult to discern. We present a general approach for rapidly mapping sequences directly to their energies on a pre-specified rigid backbone, an important sub-problem in computational protein design and in some methods for protein structure prediction. The cluster expansion (CE) method that we employ can, in principle, be extended to model any computable or measurable protein property directly as a function of sequence. Here we show how CE can be applied to the problem of computational protein design, and use it to derive excellent approximations of physical potentials. The approach provides several attractive advantages. First, following a one-time derivation of a CE expansion, the amount of time necessary to evaluate the energy of a sequence adopting a specified backbone conformation is reduced by a factor of 10(7) compared to standard full-atom methods for the same task. Second, the agreement between two full-atom methods that we tested and their CE sequence-based expressions is very high (root mean square deviation 1.1-4.7 kcal/mol, R2 = 0.7-1.0). Third, the functional form of the CE energy expression is such that individual terms of the expansion have clear physical interpretations. We derived expressions for the energies of three classic protein design targets-a coiled coil, a zinc finger, and a WW domain-as functions of sequence, and examined the most significant terms. Single-residue and residue-pair interactions are sufficient to accurately capture the energetics of the dimeric coiled coil, whereas higher-order contributions are important for the two more globular folds. For the task of designing novel zinc-finger sequences, a CE-derived energy function provides significantly better solutions than a standard design protocol, in comparable computation time. Given these advantages, CE is likely to find many uses in computational structural modeling.

Studies in animals have documented that, compared with glucose, dietary fructose induces dyslipidemia and insulin resistance. To assess the relative effects of these dietary sugars during sustained consumption in humans, overweight and obese subjects consumed glucose- or fructose-sweetened beverages providing 25% of energy requirements for 10 weeks. Although both groups exhibited similar weight gain during the intervention, visceral adipose volume was significantly increased only in subjects consuming fructose. Fasting plasma triglyceride concentrations increased by approximately 10% during 10 weeks of glucose consumption but not after fructose consumption. In contrast, hepatic de novo lipogenesis (DNL) and the 23-hour postprandial triglyceride AUC were increased specifically during fructose consumption. Similarly, markers of altered lipid metabolism and lipoprotein remodeling, including fasting apoB, LDL, small dense LDL, oxidized LDL, and postprandial concentrations of remnant-like particle-triglyceride and -cholesterol significantly increased during fructose but not glucose consumption. In addition, fasting plasma glucose and insulin levels increased and insulin sensitivity decreased in subjects consuming fructose but not in those consuming glucose. These data suggest that dietary fructose specifically increases DNL, promotes dyslipidemia, decreases insulin sensitivity, and increases visceral adiposity in overweight/obese adults.

We present Multiscale Vision Transformers (MViT) for video and image recognition, by connecting the seminal idea of multiscale feature hierarchies with transformer models. Multiscale Transformers have several channel-resolution scale stages. Starting from the input resolution and a small channel dimension, the stages hierarchically expand the channel capacity while reducing the spatial resolution. This creates a multiscale pyramid of features with early layers operating at high spatial resolution to model simple low-level visual information, and deeper layers at spatially coarse, but complex, high-dimensional features. We evaluate this fundamental architectural prior for modeling the dense nature of visual signals for a variety of video recognition tasks where it outperforms concurrent vision transformers that rely on large scale external pre-training and are 5-10x more costly in computation and parameters. We further remove the temporal dimension and apply our model for image classification where it outperforms prior work on vision transformers. Code is available at: this https URL

To determine the metabolic effects of grapefruit juice consumption we established a model in which C57Bl/6 mice drank 25-50% sweetened GFJ, clarified of larger insoluble particles by centrifugation (cGFJ), ad libitum as their sole source of liquid or isocaloric and sweetened water. cGFJ and control groups consumed similar amounts of liquids and calories. Mice fed a high-fat diet and cGFJ experienced a 18.4% decrease in weight, a 13-17% decrease in fasting blood glucose, a three-fold decrease in fasting serum insulin, and a 38% decrease in liver triacylglycerol values, compared to controls. Mice fed a low-fat diet that drank cGFJ experienced a two-fold decrease in fasting insulin, but not the other outcomes observed with the high-fat diet. cGFJ consumption decreased blood glucose to a similar extent as the commonly used anti-diabetic drug metformin. Introduction of cGFJ after onset of diet-induced obesity also reduced weight and blood glucose. A bioactive compound in cGFJ, naringin, reduced blood glucose and improved insulin tolerance, but did not ameliorate weight gain. These data from a well-controlled animal study indicate that GFJ contains more than one health-promoting neutraceutical, and warrant further studies of GFJ effects in the context of obesity and/or the western diet.

Levels of polypeptides participating in the photosynthetic light and dark reactions have been measured during fruit ripening in tomato. Photosynthetic proteins were identified by Western blot analysis with heterologous antibodies. The concentrations of proteins of photosystem (PS) I (14 kilodaltons), of PSII (47-kilodalton reaction center protein, 32-kilodalton ;Q(B) binding' protein and light harvesting complex proteins), of the photosynthetic electron transport chain (ferredoxin-NADP-oxidoreductase and plastocyanin), and of the stroma (ribulose-1,5-bisphosphate carboxylase) decrease during the ripening process. The 32-kilodalton protein and plastocyanin were detectable in pericarp protein preparations of ripe tomato fruits. Absorbance difference spectrophotometry provided information on the relative concentrations of PSII and PSI reaction centers in leaf and green fruit tissue of tomato. These results indicate that green fruit pericarp of tomato is photosynthetically active. Photosynthetic activity decreases during chloroplast/chromoplast differentiation. This is consistent with changes that occur at the transcript level of photosynthesis-specific proteins during the differentiation process.