Gene discovery efforts in autism spectrum disorder have identified heterozygous defects in chromatin remodeller genes, the 'readers, writers and erasers' of methyl marks on chromatin, as major contributors to this disease. Despite this advance, a convergent aetiology between these defects and aberrant chromatin architecture or gene expression has remained elusive. Recently, data have begun to emerge that chromatin remodellers also function directly on the cytoskeleton. Strongly associated with autism spectrum disorder, the SETD2 histone methyltransferase for example, has now been shown to directly methylate microtubules of the mitotic spindle. However, whether microtubule methylation occurs in post-mitotic cells, for example on the neuronal cytoskeleton, is not known. We found the SETD2 α-tubulin lysine 40 trimethyl mark occurs on microtubules in the brain and in primary neurons in culture, and that the SETD2 C-terminal SRI domain is required for binding and methylation of α-tubulin. A CRISPR knock-in of a pathogenic SRI domain mutation (Setd2SRI) that disables microtubule methylation revealed at least one wild-type allele was required in mice for survival, and while viable, heterozygous Setd2SRI/wtmice exhibited an anxiety-like phenotype. Finally, whereas RNA-sequencing (RNA-seq) and chromatin immunoprecipitation-sequencing (ChIP-seq) showed no concomitant changes in chromatin methylation or gene expression in Setd2SRI/wtmice, primary neurons exhibited structural deficits in axon length and dendritic arborization. These data provide the first demonstration that microtubules of neurons are methylated, and reveals a heterozygous chromatin remodeller defect that specifically disables microtubule methylation is sufficient to drive an autism-associated phenotype.

The rapid popularity of smartphones has led to a growing research interest in human activity recognition (HAR) with the mobile devices. Accelerometer is the most commonly used sensor of smartphone for HAR. Many supervised HAR methods have been developed. However, it is very difficult to collect the annotated or labeled training data for HAR. So, developing of effective unsupervised methods for HAR is very necessary. The accuracy of an unsupervised method, such as clustering, can be greatly affected by the similarity or distance measures, because the learning process of clustering method is completely depending on the similarity between objects. Although Euclidean distance measure is commonly used in unsupervised activity recognition, it is not suitable for measuring distance when the number of features is very large, which is usually the case in HAR. Jaccard distance is a distance measure based on mutual information theory and can better represent the differences between nonnegative feature vectors than Euclidean distance. It can also work well with a large number of features. In this work, the Jaccard distance measure is applied to HAR for the first time. In the experiments, the results of the Jaccard distance measure and the Euclidean distance measure are compared, using three different feature extraction methods which include time-domain, frequency-domain and mixed-domain feature extractions. To comprehensively analyze the experimental results, two different evaluation methods are used: (a) C-Index before clustering, (b) FM-index after using five different clustering methods which are Spectral Cluster, Single-Linkage, Ward-Linkage, Average-Linkage, and K-Medoids. Experiments show that, almost for every combination of the feature extraction methods and the evaluation methods, the Jaccard distance measure is consistently better than the Euclidean distance measure for unsupervised HAR.

In the last few years, research on human activity recognition using the built-in sensors of smartphones instead of the body-worn sensors has received much attention. Accelerometer is the most commonly used sensor of smartphone for the application. An important step in activity recognition is feature extraction from the raw acceleration data. In this work, a novel feature extraction method which considers both the distribution and the rate of change of the raw acceleration data is proposed. The raw time series liner acceleration data was collected by a smartphone application developed by ourselves. The proposed feature extraction method is compared with a previously proposed statistics-based feature extraction method using two evaluation methods: (a) distance matrix before clustering, (b) ARI and FM-index after clustering using MCODE. Both results show that the newly proposed feature extraction method is more effective for daily activity recognition than the previously proposed method.

In research of the copyright protection of digital text, according to the shortage of poor robustness and weakness of attack resistance for the existing text watermark, Chinese machine code and information gain strategy have been introduced into text zero-watermark scheme, which puts forward the text zero-watermark based on the Chinese Edit Distance, constructs the construction and detection process model, and designs the core algorithm. First, we introduce information gain to compute the weighting terms in paragraph to select the text feature. Then, we use Chinese machine code to express the text feature formally. Last, we compute the edit distance between the characteristic words in every paragraph and mark the commutation position to form an edit distance matrix to complete text zero-watermark construction. In detection stage, we analyze the edit distance and commutation position to reflect the distortion status. In order to validate effectiveness and feasibility of the scheme, simulation experiments are provided. The results indicated that the scheme has good robustness and strong ability against attack, especially, for the attack toward sentence type conversion and add & delete words, it has more obvious advantages.

In research of copyright protection of digital text, according to the text watermark robustness and weakness of attack resistance, the irregular surface texture mapping strategy has been introduced into text zero-watermark technology, puts forward the method based on texture mapping coordinates of text zero-watermark, and constructs the construction and detection process model, and also designs the core algorithm. First, from the word level, paragraph level and text level, those levels construct the text- curved surface projection model. Then extract the center word coordinates corresponding to the plane and curved surface in the model. Last for the surface coordinates, by using the edge-vertex texture mapping and the interior point harmonic mapping, to get the texture coordinate, and then combine it with plane coordinate to build index to complete text zero-watermark construction and detection. In order to validate the watermark scheme is effectively and feasibly, simulation tests are provided. The experimental results indicate the method has good robustness and strong ability against attack, especially, for the attack toward sentence type conversion, it has more obvious advantages.

In the study of copyright protection of digital text, a Chinese text Zero-Watermark scenario based on space model was proposed. The construction and detection model of the watermark was presented, and the core algorithm of the watermark was designed in this paper. In the scenario, the three-dimensional model was constructed by using the two-dimensional coordinate of word-level and the sentence weights of sentence-level. By mapping all sentences to the three-dimensional model, the three-dimensional space modal of whole text was produced, which was the eigenvector of this paper. Therefore, the Chinese text zero-watermark was constructed. Furthermore, the effectiveness and feasibility of the algorithm was proved with algorithm simulation. The simulation results also showed that the algorithm had good robustness, especially for syntactic transfer¿Csynonyms replacement and shear attack.

The extraction of biological symbol feature decided the recognition effect in speaker identification. Because the traditional feature extraction of lips based on single static image neglected the movement of lips in speaking, a method of speaker identification in time-sequential images based on movements of lips was proposed. Firstly, feature information of each frame in time-sequential of videos was extracted. Secondly, HMM models of movements of lips of speakers were constructed and trained. Finally, compared probability between observed symbol sequence and HMM models of database to identify the speaker. The experiments indicated that the algorithm was well done in feasibility and effectiveness and had good recognition effect.

Interferon-tau (IFNT), the pregnancy recognition signal in ruminant species, is secreted by conceptus trophectoderm cells and induces expression of IFN-stimulated gene 15 (ISG15) in the uterus and corpus luteum (CL) in ewes. Expression of ISG15 in ovine CL is speculated to be through an endocrine pathway, but it is unclear whether expression of ISG15 in bovine CL is via such a pathway. In this study, CL were obtained from cows on d 16, 25, 60, 120, 180, and 270 of pregnancy, and endometrium, mammary gland, ovarian stroma, and CL were also collected from cows on d 18 of pregnancy and on d 15 and 18 of the estrous cycle. All tissue explants from d 15 of the estrous cycle were cultured in the absence or presence of 100ng/mL of recombinant bovine IFNT for 24h. The results indicated that ISG15 and conjugated proteins were expressed in CL of both cyclic and pregnant cows regardless of pregnancy status and were upregulated during early pregnancy. The mammary gland from d 18 of pregnancy did not express ISG15, but explants of the mammary gland from d 15 of the estrous cycle did express ISG15 after being treated with IFNT. However, luteal explants from d 15 of the estrous cycle did not express ISG15 after being cultured for 24h. In conclusion, ISG15 expression is upregulated in the bovine CL during early pregnancy. Interestingly, cultured CL cells do not respond to IFNT, suggesting that the pregnancy-dependent stimulation of ISG15 expression is controlled by something other than IFNT in the bloodstream.