Many conserved noncoding sequences function as transcriptional enhancers that regulate gene expression. Here, we report that protein-coding DNA also frequently contains enhancers functioning at the transcriptional level. We tested the enhancer activity of

Quantization is an effective method for reducing memory footprint and inference time of Neural Networks. However, ultra low precision quantization could lead to significant degradation in model accuracy. A promising method to address this is to perform mixed-precision quantization, where more sensitive layers are kept at higher precision. However, the search space for a mixed-precision quantization is exponential in the number of layers. Recent work has proposed a novel Hessian based framework [9], with the aim of reducing this exponential search space by using second-order information. While promising, this prior work has three major limitations: (i) they only use a heuristic metric based on top Hessian eigenvalue as a measure of sensitivity and do not consider the rest of the Hessian spectrum; (ii) their approach only provides relative sensitivity of different layers and therefore requires a manual selection of the mixed-precision setting; and (iii) they do not consider mixed-precision activation quantization. Here, we present HAWQ-V2 which addresses these shortcomings. For (i), we theoretically prove that the right sensitivity metric is the average Hessian trace, instead of just top Hessian eigenvalue. For (ii), we develop a Pareto frontier based method for automatic bit precision selection of different layers without any manual intervention. For (iii), we develop the first Hessian based analysis for mixed-precision activation quantization, which is very beneficial for object detection. We show that HAWQ-V2 achieves new state-of-the-art results for a wide range of tasks. In particular, we present quantization results for InceptionV3 (7.57MB with 75.98% accuracy), ResNet50 (7.99MB with 75.92% accuracy), and SqueezeNext (1MB with 68.68% accuracy), all without any manual bit selection. Furthermore, we present results for object detection on Microsoft COCO, where we achieve 2.6 higher mAP than direct uniform quantization and 1.6 higher mAP than the recently proposed method of FQN, with a smaller model size of 17.9MB.

Background: Cryoablation is one of the local therapies for hepatocellular carcinoma (HCC), but its safety and effect has not been studied in patients with Child class A or B and Barcelona Clinic Liver Cancer (BCLC) stage C HCC. Metastasis-associated in colon cancer-1 (MACC1) overexpression has been associated with poor prognosis of HCC, but its predictive value to post-cryoablation outcomes remains unknown in patients with BCLC stage C HCC.Methods: This study assessed the safety and outcomes of cryoablation measured by time to progression (TTP) and overall survival (OS), and predictive value of MACC1 mRNA and protein overexpression in tumorous tissue to post-cryoablation outcomes in 120 advanced HCC patients with child-pugh class A or B by quantitative polymerase chain reaction and immunohistochemical staining. The potenial correlation of MACC1 and c-Met expression to tumor cell proliferation and apoptosis was also analyzed.Results: The cryoablation in patients with advanced unresectable HCC resulted in a median TTP and OS of 5.5 (4.2- 6.7) months and 10.5 (9.0-12.0) months, respectively and no significant complications, comparable to the historical report for RFA therapy. The MACC1 mRNA and nuclear protein expression was significantly increased in tumorous tissues in these patients than that in normal liver tissue controls. Higher expression of MACC1 mRNA and nuclear protein in tumorous tissues in these patients was associated with shorter post cryoablation median TTP and OS than that with lower MACC1 expression.Conclusions: Cryoablation is a safe and effective therapeutic option for patients with advanced HCC and Child-pugh class A or B cirrhosis; and a higher intratumoral expression of MACC1 or nuclear translocation predicts poor outcomes of cryotherapy in these patients. © 2013 Yang et al; licensee BioMed Central Ltd.

Human immunodeficiency virus-1 (HIV-1) infection disrupts the intestinal immune system, leading to microbial translocation and systemic immune activation. We investigated the impact of HIV-1 infection on the intestinal microbiome and its association with mucosal T-cell and dendritic cell (DC) frequency and activation, as well as with levels of systemic T-cell activation, inflammation, and microbial translocation. Bacterial 16S ribosomal DNA sequencing was performed on colon biopsies and fecal samples from subjects with chronic, untreated HIV-1 infection and uninfected control subjects. Colon biopsies of HIV-1-infected subjects had increased abundances of Proteobacteria and decreased abundances of Firmicutes compared with uninfected donors. Furthermore at the genus level, a significant increase in Prevotella and decrease in Bacteroides was observed in HIV-1-infected subjects, indicating a disruption in the Bacteroidetes bacterial community structure. This HIV-1-associated increase in Prevotella abundance was associated with increased numbers of activated colonic T cells and myeloid DCs. Principal coordinates analysis demonstrated an HIV-1-related change in the microbiome that was associated with increased mucosal cellular immune activation, microbial translocation, and blood T-cell activation. These observations suggest that an important relationship exists between altered mucosal bacterial communities and intestinal inflammation during chronic HIV-1 infection.