New algorithms for automated phylogenetic reconstruction using artificial intelligence and data mining techniques

Academic and industrial research in the late 90s have brought about an exponential explosion of DNA sequence data. Automated expert systems are being created to help biologists to extract patterns, trends and links from this ever-deepening ocean of information. Two such systems aimed on retrieving and subsequently utilizing phylogenetically relevant information have been developed in this dissertation, the major objective of which was to automate the often difficult and confusing phylogenetic reconstruction process. ^ Popular phylogenetic reconstruction methods, such as distance-based methods, attempt to find an optimal tree topology (that reflects the relationships among related sequences and their evolutionary history) by searching through the topology space. Various compromises between the fast (but incomplete) and exhaustive (but computationally prohibitive) search heuristics have been suggested. An intelligent compromise algorithm that relies on a flexible “beam” search principle from the Artificial Intelligence domain and uses the pre-computed local topology reliability information to adjust the beam search space continuously is described in the second chapter of this dissertation. ^ However, sometimes even a (virtually) complete distance-based method is inferior to the significantly more elaborate (and computationally expensive) maximum likelihood (ML) method. In fact, depending on the nature of the sequence data in question either method might prove to be superior. Therefore, it is difficult (even for an expert) to tell a priori which phylogenetic reconstruction method—distance-based, ML or maybe maximum parsimony (MP)—should be chosen for any particular data set. ^ A number of factors, often hidden, influence the performance of a method. For example, it is generally understood that for a phylogenetically “difficult” data set more sophisticated methods (e.g., ML) tend to be more effective and thus should be chosen. However, it is the interplay of many factors that one needs to consider in order to avoid choosing an inferior method (potentially a costly mistake, both in terms of computational expenses and in terms of reconstruction accuracy.) ^ Chapter III of this dissertation details a phylogenetic reconstruction expert system that selects a superior proper method automatically. It uses a classifier (a Decision Tree-inducing algorithm) to map a new data set to the proper phylogenetic reconstruction method. ^

The Directional Distance Function and Measurement of Super-Efficiency: An Application to Airlines Data

Lovell and Rouse (LR) have recently proposed a modification of the standard DEA model that overcomes the infeasibility problem often encountered in computing super-efficiency. In the LR procedure one appropriately scales up the observed input vector (scale down the output vector) of the relevant super-efficient firm thereby usually creating its inefficient surrogate. An alternative procedure proposed in this paper uses the directional distance function introduced by Chambers, Chung, and Färe and the resulting Nerlove-Luenberger (NL) measure of super-efficiency. The fact that the directional distance function combines features of both an input-oriented and an output-oriented model, generally leads to a more complete ranking of the observations than either of the oriented models. An added advantage of this approach is that the NL super-efficiency measure is unique and does not depend on any arbitrary choice of a scaling parameter. A data set on international airlines from Coelli, Perelman, and Griffel-Tatje (2002) is utilized in an illustrative empirical application.

Determinants and Effects of Maturity Mismatches in Emerging Markets: Evidence from Bank Level Data

Despite the extensive work on currency mismatches, research on the determinants and effects of maturity mismatches is scarce. In this paper I show that emerging market maturity mismatches are negatively affected by capital inflows and price volatilities. Furthermore, I find that banks with low maturity mismatches are more profitable during crisis periods but less profitable otherwise. The later result implies that banks face a tradeoff between higher returns and risk, hence channeling short term capital into long term loans is caused by cronyism and implicit guarantees rather than the depth of the financial market. The positive relationship between maturity mismatches and price volatility, on the other hand, shows that the banks of countries with high exchange rate and interest rate volatilities can not, or choose not to hedge themselves. These results follow from a panel regression on a data set I constructed by merging bank level data with aggregate data. This is advantageous over traditional studies which focus only on aggregate data.

Collection of data on plant height along the plant diversity gradient in the Jena Experiment (Main Experiment, time series since 2002)

This data set contains a time series of plant height measurements (vegetative and reproductive) from the main experiment plots of a large grassland biodiversity experiment (the Jena Experiment; see further details below). In addition, data on species specific plant heights for the main experiment are available from 2002. In the main experiment, 82 grassland plots of 20 x 20 m were established from a pool of 60 species belonging to four functional groups (grasses, legumes, tall and small herbs). In May 2002, varying numbers of plant species from this species pool were sown into the plots to create a gradient of plant species richness (1, 2, 4, 8, 16 and 60 species) and functional richness (1, 2, 3, 4 functional groups). Plots were maintained by bi-annual weeding and mowing. 1. Plant height was recorded, generally, twice a year just before biomass harvest (during peak standing biomass in late May and in late August). Methodologies of measuring height have varied somewhat over the years. In earlier year the streched plant height was measured, while in later years the standing height without streching the plant was measured. Vegetative height was measured either as the height of the highest leaf or as the length of the main axis of non-flowering plants. Regenerating height was measured either as the height of the highest flower on a plant or as the height of the main axis of flowering. Sampled plants were either randomly selected in the core area of plots or along transects in defined distances. For details refer to the description of individual years. Starting in 2006, also the plots of the management experiment, that altered mowing frequency and fertilized subplots (see further details in the general description of the Jena Experiment) were sampled. 2. Species specific plant height was recorded two times in 2002: in late July (vegetative height) and just before biomass harvest during peak standing biomass in late August (vegetative and regenerative height). For each plot and each sown species in the species pool, 3 plant individuals (if present) from the central area of the plots were randomly selected and used to measure vegetative height (non-flowering indviduals) and regenerative height (flowering individuals) as stretched height. Provided are the means over the three measuremnts per plant species per plot.

Collection of data on soil carbon (particulate and dissolved) in the Jena Experiment (Main Experiment, time series since 2002)

This data set contains three time series of measurements of soil carbon (particular and dissolved) from the main experiment plots of a large grassland biodiversity experiment (the Jena Experiment; see further details below). In the main experiment, 82 grassland plots of 20 x 20 m were established from a pool of 60 species belonging to four functional groups (grasses, legumes, tall and small herbs). In May 2002, varying numbers of plant species from this species pool were sown into the plots to create a gradient of plant species richness (1, 2, 4, 8, 16 and 60 species) and functional richness (1, 2, 3, 4 functional groups). Plots were maintained by bi-annual weeding and mowing. 1. Particulate soil carbon: Stratified soil sampling was performed every two years since before sowing in April 2002 and was repeated in April 2004, 2006 and 2008 to a depth of 30 cm segmented to a depth resolution of 5 cm giving six depth subsamples per core. Total carbon concentration was analyzed on ball-milled subsamples by an elemental analyzer at 1150°C. Inorganic carbon concentration was measured by elemental analysis at 1150°C after removal of organic carbon for 16 h at 450°C in a muffle furnace. Organic carbon concentration was calculated as the difference between both measurements of total and inorganic carbon. 2. Particulate soil carbon (high intensity sampling): In one block of the Jena Experiment soil samples were taken to a depth of 1 m (segmented to a depth resolution of 5 cm giving 20 depth subsamples per core) with three replicates per block ever 5 years starting before sowing in April 2002. Samples were processed as for the more frequent sampling. 3. Dissolved organic carbon: Suction plates installed on the field site in 10, 20, 30 and 60 cm depth were used to sample soil pore water. Cumulative soil solution was sampled biweekly and analyzed for dissolved organic carbon concentration by a high TOC elemental analyzer. Annual mean values of DOC are provided.

Soil, snow, weather, and sub-surface storage data from a mountain catchment in the rain-snow transition zone

A comprehensive hydroclimatic data set is presented for the 2011 water year to improve understanding of hydrologic processes in the rain-snow transition zone. This type of dataset is extremely rare in scientific literature because of the quality and quantity of soil depth, soil texture, soil moisture, and soil temperature data. Standard meteorological and snow cover data for the entire 2011 water year are included, which include several rain-on-snow events. Surface soil textures and soil depths from 57 points are presented as well as soil texture profiles from 14 points. Meteorological data include continuous hourly shielded, unshielded, and wind corrected precipitation, wind speed, air temperature, relative humidity, dew point temperature, and incoming solar and thermal radiation data. Sub-surface data included are hourly soil moisture data from multiple depths from 7 soil profiles within the catchment, and soil temperatures from multiple depths from 2 soil profiles. Hydrologic response data include hourly stream discharge from the catchment outlet weir, continuous snow depths from one location, intermittent snow depths from 5 locations, and snow depth and density data from ten weekly snow surveys. Though it represents only a single water year, the presentation of both above and below ground hydrologic condition makes it one of the most detailed and complete hydro-climatic datasets from the climatically sensitive rain-snow transition zone for a wide range of modeling and descriptive studies.

Part of the global DOC versus AOU (dissolved organic carbon/apparent oxygen utilization) data compilation, BATS

The contributions of total organic carbon and nitrogen to elemental cycling in the surface layer of the Sargasso Sea are evaluated using a 5-yr time-series data set (1994-1998). Surface-layer total organic carbon (TOC) and total organic nitrogen (TON) concentrations ranged from 60 to 70 µM C and 4 to 5.5 µM N seasonally, resulting in a mean C : N molar ratio of 14.4±2.2. The highest surface concentrations varied little during individual summer periods, indicating that net TOC production ceased during the highly oligotrophic summer season. Winter overturn and mixing of the water column were both the cause of concentration reductions and the trigger for net TOC production each year following nutrient entrainment and subsequent new production. The net production of TOC varied with the maximum in the winter mixed-layer depth (MLD), with greater mixing supporting the greatest net production of TOC. In winter 1995, the TOC stock increased by 1.4 mol C/m**2 in response to maximum mixing depths of 260 m. In subsequent years experiencing shallower maxima in MLD (<220 m), TOC stocks increased <0.7 mol C/m**2. Overturn of the water column served to export TOC to depth (>100 m), with the amount exported dependent on the depth of mixing (total export ranged from 0.4 to 1.4 mol C/m**2/yr). The exported TOC was comprised both of material resident in the surface layer during late summer (resident TOC) and material newly produced during the spring bloom period (fresh TOC). Export of resident TOC ranged from 0.5 to 0.8 mol C/m**2/yr, covarying with the maximum winter MLD. Export of fresh TOC varied from nil to 0.8 mol C/m**2/yr. Fresh TOC was exported only after a threshold maximum winter MLD of ~200 m was reached. In years with shallower mixing, fresh TOC export and net TOC production in the surface layer were greatly reduced. The decay rates of the exported TOC also covaried with maximum MLD. The year with deepest mixing resulted in the highest export and the highest decay rate (0.003 1/d) while shallow and low export resulted in low decay rates (0.0002 1/d), likely a consequence of the quality of material exported. The exported TOC supported oxygen utilization at dC : dO2 molar ratios ranging from 0.17 when TOC export was low to 0.47 when it was high. We estimate that exported TOC drove 15-41% of the annual oxygen utilization rates in the 100-400 m depth range. Finally, there was a lack of variability in the surface-layer TON signal during summer. The lack of a summer signal for net TON production suggests a small role for N2 fixation at the site. We hypothesize that if N2 fixation is responsible for elevated N : P ratios in the main thermocline of the Sargasso Sea, then the process must take place south of Bermuda and the signal transported north with the Gulf Stream system.

Collection of data on particulate and dissolved soil phosphorus in the Jena Experiment (Main Experiment, time series since 2002)

This data set contains two time series of measurements of dissolved phosphorus (organic, inorganic and total with a biweekly resolution) and dissolved inorganic phosphorus with a seasonal resolution. In addition, data on phosphorus from soil samples measured in 2007 and fractionated by different acid-extrations (Hedley fractions) are provided. All data measured at the main experiment plots of a large grassland biodiversity experiment (the Jena Experiment; see further details below). In the main experiment, 82 grassland plots of 20 x 20 m were established from a pool of 60 species belonging to four functional groups (grasses, legumes, tall and small herbs). In May 2002, varying numbers of plant species from this species pool were sown into the plots to create a gradient of plant species richness (1, 2, 4, 8, 16 and 60 species) and functional richness (1, 2, 3, 4 functional groups). Plots were maintained by bi-annual weeding and mowing. 1. Dissolved phosphorus in soil solution: Suction plates installed on the field site in 10, 20, 30 and 60 cm depth were used to sample soil pore water. Cumulatively extracted soil solution was collected every two weeks from October 2002 to May 2006. The biweekly samples from 2002, 2003 and 2004 were analyzed for dissolved organic phosphorus (DOP), dissolved inorganic phosphorus (PO4P) and dissolved total phosphorus (TDP) by Continuous Flow Analyzer (CFA SAN ++, SKALAR [Breda, The Netherlands]). 2. Seasonal values of dissolved inorganic phosphorus in soil solution were calculated as volume-weighted mean values of the biweekly measurements (spring = March to May, summer = June to August, fall = September to November, winter = December to February). 3. Phosphorus fractions in soil: Five independent soil samples per plot were taken in a depth of 0-15 cm using a soil corer with an inner diameter of 1 cm. The five samples per plot were combined to one composite sample per plot. A four-step sequential P fractionation (Hedley fractions) was applied and concentrations of P fractions in soil were measured photometrically (molybdenum blue-reactive P) with a Continuous Flow Analyzer (Bran&Luebbe, Germany).