Learning Temporal Causal Sequence Relationships from Real-Time Time-Series

We aim to mine temporal causal sequences that explain observed events (consequents) in time-series traces. Causal explanations of key events in a time-series have applications in design debugging, anomaly detection, planning, root-cause analysis and many more. We make use of decision trees and interval arithmetic to mine sequences that explain defining events in the time-series. We propose modified decision tree construction metrics to handle the non-determinism introduced by the temporal dimension. The mined sequences are expressed in a readable temporal logic language that is easy to interpret. The application of the proposed methodology is illustrated through various examples.

Robust dynamical invariants in sequential neural activity

By studying different sources of temporal variability in central pattern generator circuits, in this paper we unveil distinct aspects of the instantaneous balance between flexibility and robustness in sequential dynamics –a property that characterizes many systems that display neural rhythms. The level of irregularity and coordination was characterized using intrinsic time references and intervals in long recordings of the pyloric central pattern generator. The analysis demonstrated strong robustness of transient dynamics in keeping not only the activation sequences but also specific cycle-by-cycle temporal relationships in the form of dynamical invariants. The rich dynamics of neurons and connections balance flexibility and coordination to readily negotiate the interactions between neurons and produce the resultant rhythm. In particular, two dynamical invariants were identified between time intervals that build the sequence, existing even outside steady states. We suggest that invariant temporal sequence relationships could be present in other networks, including those related to brain rhythms, and underlie rhythm programming and functionality.

Graph regularized, semi-supervised learning improves annotation of de novo transcriptomes

We present a new method, GRASS, for improving an initial annotation of de novo transcriptomes. GRASS makes the shared-sequence relationships between assembled contigs explicit in the form of a graph, and applies an algorithm that performs label propagation to transfer annotations between related contigs and modifies the graph topology iteratively. We demonstrate that GRASS increases the completeness and accuracy of the initial annotation, allows for improved differential analysis, and is very efficient, typically taking 10s of minutes.

Can DNA-based ecosystem assessments quantify species abundance? Testing primer bias and biomass - sequence relationships with an innovative metabarcoding protocol

Metabarcoding is an emerging genetic tool to rapidly assess biodiversity in ecosystems. It involves high-throughput sequencing of a standard gene from an environmental sample and comparison to a reference database. However, no consensus has emerged regarding laboratory pipelines to screen species diversity and infer species abundances from environmental samples. In particular, the effect of primer bias and the detection limit for specimens with a low biomass has not been systematically examined, when processing samples in bulk. We developed and tested a DNA metabarcoding protocol that utilises the standard cytochrome c oxidase subunit I (COI) barcoding fragment to detect freshwater macroinvertebrate taxa. DNA was extracted in bulk, amplified in a single PCR step, and purified, and the libraries were directly sequenced in two independent MiSeq runs (300-bp paired-end reads). Specifically, we assessed the influence of specimen biomass on sequence read abundance by sequencing 31 specimens of a stonefly species with known haplotypes spanning three orders of magnitude in biomass (experiment I). Then, we tested the recovery of 52 different freshwater invertebrate taxa of similar biomass using the same standard barcoding primers (experiment II). Each experiment was replicated ten times to maximise statistical power. The results of both experiments were consistent across replicates. We found a distinct positive correlation between species biomass and resulting numbers of MiSeq reads. Furthermore, we reliably recovered 83% of the 52 taxa used to test primer bias. However, sequence abundance varied by four orders of magnitudes between taxa despite the use of similar amounts of biomass. Our metabarcoding approach yielded reliable results for high-throughput assessments. However, the results indicated that primer efficiency is highly species-specific, which would prevent straightforward assessments of species abundance and biomass in a sample. Thus, PCR-based metabarcoding assessments of biodiversity should rely on presence-absence metrics.

Can DNA-based ecosystem assessments quantify species abundance? Testing primer bias and biomass - sequence relationships with an innovative metabarcoding protocol

Metabarcoding is an emerging genetic tool to rapidly assess biodiversity in ecosystems. It involves high-throughput sequencing of a standard gene from an environmental sample and comparison to a reference database. However, no consensus has emerged regarding laboratory pipelines to screen species diversity and infer species abundances from environmental samples. In particular, the effect of primer bias and the detection limit for specimens with a low biomass has not been systematically examined, when processing samples in bulk. We developed and tested a DNA metabarcoding protocol that utilises the standard cytochrome c oxidase subunit I (COI) barcoding fragment to detect freshwater macroinvertebrate taxa. DNA was extracted in bulk, amplified in a single PCR step, and purified, and the libraries were directly sequenced in two independent MiSeq runs (300-bp paired-end reads). Specifically, we assessed the influence of specimen biomass on sequence read abundance by sequencing 31 specimens of a stonefly species with known haplotypes spanning three orders of magnitude in biomass (experiment I). Then, we tested the recovery of 52 different freshwater invertebrate taxa of similar biomass using the same standard barcoding primers (experiment II). Each experiment was replicated ten times to maximise statistical power. The results of both experiments were consistent across replicates. We found a distinct positive correlation between species biomass and resulting numbers of MiSeq reads. Furthermore, we reliably recovered 83% of the 52 taxa used to test primer bias. However, sequence abundance varied by four orders of magnitudes between taxa despite the use of similar amounts of biomass. Our metabarcoding approach yielded reliable results for high-throughput assessments. However, the results indicated that primer efficiency is highly species-specific, which would prevent straightforward assessments of species abundance and biomass in a sample. Thus, PCR-based metabarcoding assessments of biodiversity should rely on presence-absence metrics.

Real-Time Shop-Floor Production Performance Analysis Method for the Internet of Manufacturing Things

Typical challenges that manufacturing enterprises are facing now are compounded by lack of timely, accurate, and consistent information of manufacturing resources. As a result, it is difficult to analyze the real-time production performance for the shop-floor. In this paper, the definition and overall architecture of the internet of manufacturing things is presented to provide a new paradigm by extending the techniques of internet of things (IoT) to manufacturing field. Under this architecture, the real-time primitive events which occurred at different manufacturing things such as operators, machines, pallets, key materials, and so forth can be easily sensed. Based on these distributed primitive events, a critical event model is established to automatically analyze the real-time production performance. Here, the up-level production performance analysis is regarded as a series of critical events, and the real-time value of each critical event can be easily calculated according to the logical and sequence relationships among these multilevel events. Finally, a case study is used to illustrate how to apply the designed methods to analyze the real-time production performance.