Exploring the Whole Rashomon Set of Sparse Decision Trees.

In any given machine learning problem, there might be many models that explain the data almost equally well. However, most learning algorithms return only one of these models, leaving practitioners with no practical way to explore alternative models that might have desirable properties beyond what could be expressed by a loss function. The Rashomon set is the set of these all almost-optimal models. Rashomon sets can be large in size and complicated in structure, particularly for highly nonlinear function classes that allow complex interaction terms, such as decision trees. We provide the first technique for completely enumerating the Rashomon set for sparse decision trees; in fact, our work provides the first complete enumeration of any Rashomon set for a non-trivial problem with a highly nonlinear discrete function class. This allows the user an unprecedented level of control over model choice among all models that are approximately equally good. We represent the Rashomon set in a specialized data structure that supports efficient querying and sampling. We show three applications of the Rashomon set: 1) it can be used to study variable importance for the set of almost-optimal trees (as opposed to a single tree), 2) the Rashomon set for accuracy enables enumeration of the Rashomon sets for balanced accuracy and F1-score, and 3) the Rashomon set for a full dataset can be used to produce Rashomon sets constructed with only subsets of the data set. Thus, we are able to examine Rashomon sets across problems with a new lens, enabling users to choose models rather than be at the mercy of an algorithm that produces only a single model.

DNA microstructure based on self-assembly of 4-sticky-end Holliday junctions in aqueous solution.

Liverwort-like DNA microscale structures consist of 4-sticky-end Holiday junctions as DNA bricks that can be used in nanotechnology and nanobiotechnology to direct the self-assembly of nanomachines as well as DNA assembly. Previously it has not been possible to obtain such DNA microscale structural forms, but herein we report construction of a mesh-like material made up of 4 strands of 40-base DNA. Advanced bioimaging techniques such as fluorescence correlation spectroscopy (FCS), laser scanning microscopy (LSM), and atomic force microscopy (AFM) help us as ultrasensitive detection tools for examing structures in solutions. Combinations of these techniques allow us to survey various chemical conditions of materials and solutions.

Specifically associated PCR products probed by coincident detection of two-color cross-correlated fluorescence intensities in human gene polymorphisms of methylene tetrahydrofolate reductase at site C677T: a novel measurement approach without follow-up mathematical analysis.

Whole blood samples of known methylene tetrahydrofolate reductase (MTHFR) genotypes from 24 individuals were examined at site C677T. Their amplified DNA products were assessed by two-color fluorescence cross-correlation measurements and agarose gel electrophoresis/capillary gel electrophoresis. DNA subpopulations were identified which were not associated with the proper genotype by primer combinations and cycling conditions called multiplexes. We confirmed that DNA analysis by two-color fluorescence cross-correlation measurements allowed the detection of fluorescence signals specifically associated with the proper genotypes in a mixture of amplified nontarget DNA molecules without DNA sizing. The measurement approach does not require complex, follow-up mathematical analysis and is applicable to any single nucleotide polymorphisms. The simple immunogenetic model showed how the approach works to reveal specific DNA target by preventing detection of nontarget DNA. Under those experimental conditions, a new ultrasensitive, and specific method for clinical immunologists is born.

DNA measurements by using fluorescence correlation spectroscopy and two-color fluorescence cross correlation spectroscopy.

FCS and FCCS measurements provide two important analytical parameters, the average number of molecules in the detection area and the translational diffusion constant of the molecules at the single molecule level. Considering these properties, FCS and FCCS have been applied to analysis of the cellular environment and dynamic processes of molecules in the living cell. More recently, a systematic approach for the analysis of macromolecule complex formation has focused on the new field of single molecule detection in the post-genome era. In this work, we tested the sensitivities of FCS and FCCS based on the distance between fluorophores in DNA as a model macromolecule complex. The results show that FCCS is not limited by the size of the macromolecular complex even in a very small detection area.

C677T single nucleotide polymorphisms of the human methylene tetrahydrofolate reductase and specific identification : a novel strategy using two-color cross-correlation fluorescence spectroscopy.

BACKGROUND: A methylene tetrahydrofolate reductase (MTHFR) deficiency at site C677T renders the enzyme thermolabile and consequently represents a risk factor for vascular disease, neural tube defects, preeclampsia, and thrombosis. Highly specific identification techniques for genotyping are mandatory to give guidance for the diagnosis and monitoring of this deficiency. METHODS: A new approach for performing genotyping has been introduced with the identification of single nucleotide polymorphisms of the human MTHFR. It is based on PCR followed by two-color cross-correlation fluorescence spectroscopy (FCS). Experiments were carried out with green- and red-tagged allele-specific primers, which were fully compatible with the two-color fluorescence cross-correlation setup at 488 nm and 633 nm excitation wavelengths. RESULTS: The measured data of the amplification mixes (tubes) were normalized as the maximum correlation amplitude of each tube. Correlated and uncorrelated data were optically separated in the amplification mixes by their characteristic correlation times, which significantly differed from each other. The correlated data were generated in the presence of the proper mutated genotype template, whereas uncorrelated data were due to the absence of the proper genotype template. Furthermore, the specific association of the two-color fluorescence correlated signals with the target DNA was experimentally proven. Using this novel two-color cross-correlation approach, the MTHFR genotypes, which were determined in 21 clinical samples, showed concordance with methods involving a PCR-based assay with hexachloro-6-carboxy-fluorescein (HEX)- and 6-carboxy-fluorescein (FAM)-tagged allele-specific primers and a subsequent separation step with capillary electrophoresis, yet are simpler to perform. There was no evidence of a central trend of false-positive or false-negative results. We demonstrated how the novel, ultrasensitive typing system could be applied to studies where researchers are trying to perfect their assays and are often working with the unknown, or application to problematic assays in a clinical environment for those involved in molecular diagnosis. CONCLUSIONS: We present an alternative method to those commonly used in genotyping. Two-color cross-correlation FCS allows the detection of the fluorescence signals specifically associated with the heterozygous mutated, the homozygous mutated, and normal individuals, as exemplified in this study. The presence of nonspecific amplification products, which interfere with subsequent DNA analysis, could therefore highlight the need for two-color cross-correlation FCS as a means of discriminating between specific association of the fluorescence signals with the target DNA and DNA not related to the target.

Family history and risk of subarachnoid hemorrhage: a case-control study in Nagoya, Japan.

BACKGROUND AND PURPOSE: We sought to examine the relation between a family history of subarachnoid hemorrhage (SAH) and the risk of SAH by using a case-control study. METHODS: Case subjects consisted of a consecutive series of 195 patients with spontaneous SAH, aged 30 to 79 years, with aneurysms confirmed by angiography and/or CT scan. Hospital and community control subjects were identified and matched to each case by sex and age (+/-2 years). Multiple conditional logistic regression was used to calculate the odds ratio (OR) and 95% interval (CI) adjusted for potential confounders. RESULTS: Having a family member with SAH was significantly associated with an increased risk of SAH (OR, 4.0, 95% CI, 2.0 to 8.0), after adjusting for potential confounders. The risk for a positive family history of SAH was similar for men and women and was inversely related to the SAH patient's age. A maternal positive SAH history (OR, 5.4; 95% CI, 1.8 to 16.0) posed a much greater risk than a paternal positive history (OR, 3.2, 95% CI, 1.1 to 13.4). CONCLUSIONS: A positive family history of SAH was significantly and strongly associated with the risk of SAH. To prevent the onset of SAH at a younger age, much more attention should be given to individuals with any family member (first-degree relatives) suffering SAH episodes.