Disease progression subtype discovery from longitudinal EMR data with a majority of missing values and unknown initial time points.

Electronic medical records (EMR) contain a longitudinal collection of laboratory data that contains valuable phenotypic information on disease progression of a large collection of patients. These data can be potentially used in medical research or patient care; finding disease progression subtypes is a particularly important application. There are, however, two significant difficulties in utilizing this data for statistical analysis: (a) a large proportion of data is missing and (b) patients are in very different stages of disease progression and there are no well-defined start points of the time series. We present a Bayesian machine learning model that overcomes these difficulties. The method can use highly incomplete time-series measurement of varying lengths, it aligns together similar trajectories in different phases and is capable of finding consistent disease progression subtypes. We demonstrate the method on finding chronic kidney disease progression subtypes.

High density lipoprotein structural changes and drug response in lipidomic profiles following the long-term fenofibrate therapy in the FIELD substudy.

In a recent FIELD study the fenofibrate therapy surprisingly failed to achieve significant benefit over placebo in the primary endpoint of coronary heart disease events. Increased levels of atherogenic homocysteine were observed in some patients assigned to fenofibrate therapy but the molecular mechanisms behind this are poorly understood. Herein we investigated HDL lipidomic profiles associated with fenofibrate treatment and the drug-induced Hcy levels in the FIELD substudy. We found that fenofibrate leads to complex HDL compositional changes including increased apoA-II, diminishment of lysophosphatidylcholines and increase of sphingomyelins. Ethanolamine plasmalogens were diminished only in a subgroup of fenofibrate-treated patients with elevated homocysteine levels. Finally we performed molecular dynamics simulations to qualitatively reconstitute HDL particles in silico. We found that increased number of apoA-II excludes neutral lipids from HDL surface and apoA-II is more deeply buried in the lipid matrix than apoA-I. In conclusion, a detailed molecular characterization of HDL may provide surrogates for predictors of drug response and thus help identify the patients who might benefit from fenofibrate treatment.

Multivariate multi-way analysis of multi-source data.

MOTIVATION: Analysis of variance (ANOVA)-type methods are the default tool for the analysis of data with multiple covariates. These tools have been generalized to the multivariate analysis of high-throughput biological datasets, where the main challenge is the problem of small sample size and high dimensionality. However, the existing multi-way analysis methods are not designed for the currently increasingly important experiments where data is obtained from multiple sources. Common examples of such settings include integrated analysis of metabolic and gene expression profiles, or metabolic profiles from several tissues in our case, in a controlled multi-way experimental setup where disease status, medical treatment, gender and time-series are usual covariates. RESULTS: We extend the applicability area of multivariate, multi-way ANOVA-type methods to multi-source cases by introducing a novel Bayesian model. The method is capable of finding covariate-related dependencies between the sources. It assumes the measurements consist of groups of similarly behaving variables, and estimates the multivariate covariate effects and their interaction effects for the discovered groups of variables. In particular, the method partitions the effects to those shared between the sources and to source-specific ones. The method is specifically designed for datasets with small sample sizes and high dimensionality. We apply the method to a lipidomics dataset from a lung cancer study with two-way experimental setup, where measurements from several tissues with mostly distinct lipids have been taken. The method is also directly applicable to gene expression and proteomics. AVAILABILITY: An R-implementation is available at http://www.cis.hut.fi/projects/mi/software/multiWayCCA/.

Dynamical scaling exponents for polymer translocation through a nanopore.

We determine the scaling exponents of polymer translocation (PT) through a nanopore by extensive computer simulations of various microscopic models for chain lengths extending up to N=800 in some cases. We focus on the scaling of the average PT time tau approximately Nalpha and the mean-square change of the PT coordinate, approximately tbeta. We find alpha=1+2nu and beta=2/alpha for unbiased PT in two dimensions (2D) and three dimensions (3D). The relation alphabeta=2 holds for driven PT in 2D, with a crossover from alpha approximately 2nu for short chains to alpha approximately 1+nu for long chains. This crossover is, however, absent in 3D where alpha=1.42+/-0.01 and alphabeta approximately 2.2 for N approximately 40-800.

Polymer translocation through a nanopore under a pulling force.

We investigate polymer translocation through a nanopore under a pulling force using Langevin dynamics simulations. We concentrate on the influence of the chain length N and the pulling force F on the translocation time tau . The distribution of tau is symmetric and narrow for strong F . We find that tau approximately N{2} and translocation velocity v approximately N{-1} for both moderate and strong F . For infinitely wide pores, three regimes are observed for tau as a function of F . With increasing F , tau is independent of F for weak F , and then tau approximately F{-2+nu{-1}} for moderate F, where nu is the Flory exponent, which finally crosses over to tau approximately F{-1} for strong force. For narrow pores, even for moderate force tau approximately F{-1}. Finally, the waiting time, for monomer s and monomer s+1 to exit the pore, has a maximum for s close to the end of the chain, in contrast to the case where the polymer is driven by an external force within the pore.

Langevin dynamics simulations of polymer translocation through nanopores.

We investigate the dynamics of polymer translocation through a nanopore using two-dimensional Langevin dynamics simulations. In the absence of an external driving force, we consider a polymer which is initially placed in the middle of the pore and study the escape time tau(e) required for the polymer to completely exit the pore on either side. The distribution of the escape times is wide and has a long tail. We find that tau(e) scales with the chain length N as tau(e) approximately N(1+2nu), where nu is the Flory exponent. For driven translocation, we concentrate on the influence of the friction coefficient xi, the driving force E, and the length of the chain N on the translocation time tau, which is defined as the time duration between the first monomer entering the pore and the last monomer leaving the pore. For strong driving forces, the distribution of translocation times is symmetric and narrow without a long tail and tau approximately E(-1). The influence of xi depends on the ratio between the driving and frictional forces. For intermediate xi, we find a crossover scaling for tau with N from tau approximately N(2nu) for relatively short chains to tau approximately N(1+nu) for longer chains. However, for higher xi, only tau approximately N(1+nu) is observed even for short chains, and there is no crossover behavior. This result can be explained by the fact that increasing xi increases the Rouse relaxation time of the chain, in which case even relatively short chains have no time to relax during translocation. Our results are in good agreement with previous simulations based on the fluctuating bond lattice model of polymers at intermediate friction values, but reveal additional features of dependency on friction.

Polymer translocation through a nanopore under an applied external field.

We investigate the dynamics of polymer translocation through a nanopore under an externally applied field using the two-dimensional fluctuating bond model with single-segment Monte Carlo moves. We concentrate on the influence of the field strength E, length of the chain N, and length of the pore L on forced translocation. As our main result, we find a crossover scaling for the translocation time tau with the chain length from tau approximately N2nu for relatively short polymers to tau approximately N1+nu for longer chains, where nu is the Flory exponent. We demonstrate that this crossover is due to the change in the dependence of the translocation velocity v on the chain length. For relatively short chains v approximately N-nu, which crosses over to v approximately N(-1) for long polymers. The reason for this is that with increasing N there is a high density of segments near the exit of the pore, which slows down the translocation process due to slow relaxation of the chain. For the case of a long nanopore for which R parallel, the radius of gyration Rg along the pore, is smaller than the pore length, we find no clear scaling of the translocation time with the chain length. For large N, however, the asymptotic scaling tau approximately N1+nu is recovered. In this regime, tau is almost independent of L. We have previously found that for a polymer, which is initially placed in the middle of the pore, there is a minimum in the escape time for R parallel approximately L. We show here that this minimum persists for weak fields E such that EL is less than some critical value, but vanishes for large values of EL.