Robust silicon arbitrary ratio power splitters using shortcuts to adiabaticity.

We design and fabricate a series of broadband silicon arbitrary power splitters with various split ratios using shortcuts to adiabaticity. In this approach, the system evolution is designed using the decoupled system states, and the desired split ratios are guaranteed by the boundary conditions. Furthermore, the system evolutions are optimized to be as close to the adiabatic states as possible, thus enhancing the robustness to wavelength and fabrication variations. The devices are more compact then the conventional adiabatic designs. Fabricated devices show broadband response for a wide wavelength range from 1.47 to 1.62 µm and also have excellent robustness against fabrication errors across an 8-inch wafer.

RadixHap: a radix tree-based heuristic for solving the single individual haplotyping problem.

Single nucleotide polymorphism studies have recently received significant amount of attention from researchers in many life science disciplines. Previous researches indicated that a series of SNPs from the same chromosome, called haplotype, contains more information than individual SNPs. Hence, discovering ways to reconstruct reliable Single Individual Haplotypes becomes one of the core issues in the whole-genome research nowadays. However, obtaining sequence from current high-throughput sequencing technologies always contain inevitable sequencing errors and/or missing information. The SIH reconstruction problem can be formulated as bi-partitioning the input SNP fragment matrix into paternal and maternal sections to achieve minimum error correction; a problem that is proved to be NP-hard. In this study, we introduce a greedy approach, named RadixHap, to handle data sets with high error rates. The experimental results show that RadixHap can generate highly reliable results in most cases. Furthermore, the algorithm structure of RadixHap is particularly suitable for whole-genome scale data sets.

The evolutionary landscape of the Mycobacterium tuberculosis genome.

Mycobacterium tuberculosis is one of the most deadly human pathogens. The major mechanism for the adaptations of M. tuberculosis is nucleotide substitution. Previous studies have relied on the nonsynonymous-to-synonymous substitution rate (dN/dS) ratio as a measurement of selective constraint based on the assumed selective neutrality of synonymous substitutions. However, this assumption has been shown to be untrue in many cases. In this study, we used the substitution rate in intergenic regions (di) of the M. tuberculosis genome as the neutral reference, and conducted a genome-wide profiling for di, dS, and the rate of insertions/deletions (indel rate) as compared with the genome of M. canettii using a 50kb sliding window. We demonstrate significant variations in all of the three evolutionary measurements across the M. tuberculosis genome, even for regions in close vicinity. Furthermore, we identified a total of 233 genes with their dS deviating significantly from di within the same window. Interestingly, dS also varies significantly in some of the windows, indicating drastic changes in mutation rate and/or selection pressure within relatively short distances in the M. tuberculosis genome. Importantly, our results indicate that selection on synonymous substitutions is common in the M. tuberculosis genome. Therefore, the dN/dS ratio test must be applied carefully for measuring selection pressure on M. tuberculosis genes.

Using genetic algorithm in reconstructing single individual haplotype with minimum error correction.

Discovering ways to reconstruct reliable Single Individual Haplotypes (SIHs) becomes one of the core issues in the whole-genome research nowadays as previous research showed that haplotypes contain more information than individual Singular Nucleotide Polymorphisms (SNPs). Although with advances in high-throughput sequencing technologies obtaining sequence information is becoming easier in today's laboratories, obtained sequences from current technologies always contain inevitable sequence errors and missing information. The SIH reconstruction problem can be formulated as bi-partitioning the input SNP fragment matrix into paternal and maternal sections to achieve minimum error correction (MEC) time; the problem that is proved to be NP-hard. Several heuristics or greedy algorithms have already been designed and implemented to solve this problem, most of them however (1) do not have the ability to handle data sets with high error rates and/or (2) can only handle binary input matrices. In this study, we introduce a Genetic Algorithm (GA) based method, named GAHap, to reconstruct SIHs with lowest MEC times. GAHap is equipped with a well-designed fitness function to obtain better reconstruction rates. GAHap is also compared with existing methods to show its ability in generating highly reliable solutions.