Pekka Parviainen

Ancestor relations in the presence of unobserved variables

Bayesian networks (BNs) are an appealing model for causal and noncausal dependencies among a set of variables. Learning BNs from observational data is challenging due to the nonidentifiability of the network structure and model misspecification in the presence of unobserved (latent) variables. Here, we investigate the prospects of Bayesian learning of ancestor relations, including arcs, in the presence and absence of unobserved variables. An exact dynamic programming algorithm to compute the respective posterior probabilities is developed, under the complete data assumption. Our experimental results show that ancestor relations between observed variables, arcs in particular, can be learned with good power even when a majority of the involved variables are unobserved. For comparison, deduction of ancestor relations from single maximum a posteriori network structures or their Markov equivalence class appears somewhat inferior to Bayesian averaging. We also discuss some shortcomings of applying existing conditional independence test based methods for learning ancestor relations.

Temperament Clusters in a Normal Population: Implications for Health and Disease

Background The object of this study was to identify temperament patterns in the Finnish population, and to determine the relationship between these profiles and life habits, socioeconomic status, and health. Methods/Principal Findings A cluster analysis of the Temperament and Character Inventory subscales was performed on 3,761 individuals from the Northern Finland Birth Cohort 1966 and replicated on 2,097 individuals from the Cardiovascular Risk in Young Finns study. Clusters were formed using the k-means method and their relationship with 115 variables from the areas of life habits, socioeconomic status and health was examined. Results Four clusters were identified for both genders. Individuals from Cluster I are characterized by high persistence, low extravagance and disorderliness. They have healthy life habits, and lowest scores in most of the measures for psychiatric disorders. Cluster II individuals are characterized by low harm avoidance and high novelty seeking. They report the best physical capacity and highest level of income, but also high rate of divorce, smoking, and alcohol consumption. Individuals from Cluster III are not characterized by any extreme characteristic. Individuals from Cluster IV are characterized by high levels of harm avoidance, low levels of exploratory excitability and attachment, and score the lowest in most measures of health and well-being. Conclusions This study shows that the temperament subscales do not distribute randomly but have an endogenous structure, and that these patterns have strong associations to health, life events, and well-being.

Fast Zeta Transforms for Lattices with Few Irreducibles

We investigate fast algorithms for changing between the standard basis and an orthogonal basis of idempotents for Möbius algebras of finite lattices. We show that every lattice with v elements, n of which are nonzero and join-irreducible (or, by a dual result, nonzero and meet-irreducible), has arithmetic circuits of size O(vn) for computing the zeta transform and its inverse, thus enabling fast multiplication in the Möbius algebra. Furthermore, the circuit construction in fact gives optimal (up to constants) monotone circuits for several lattices of combinatorial and algebraic relevance, such as the lattice of subsets of a finite set, the lattice of set partitions of a finite set, the lattice of vector subspaces of a finite vector space, and the lattice of positive divisors of a positive integer.

Species Tree Inference Using a Mixture Model

Species tree reconstruction has been a subject of substantial research due to its central role across biology and medicine. A species tree is often reconstructed using a set of gene trees or by directly using sequence data. In either of these cases, one of the main confounding phenomena is the discordance between a species tree and a gene tree due to evolutionary events such as duplications and losses. Probabilistic methods can resolve the discordance by coestimating gene trees and the species tree but this approach poses a scalability problem for larger data sets. We present MixTreEM-DLRS: A two-phase approach for reconstructing a species tree in the presence of gene duplications and losses. In the first phase, MixTreEM, a novel structural expectation maximization algorithm based on a mixture model is used to reconstruct a set of candidate species trees, given sequence data for monocopy gene families from the genomes under study. In the second phase, PrIME-DLRS, a method based on the DLRS model (Åkerborg O, Sennblad B, Arvestad L, Lagergren J. 2009. Simultaneous Bayesian gene tree reconstruction and reconciliation analysis. Proc Natl Acad Sci U S A. 106(14):5714-5719), is used for selecting the best species tree. PrIME-DLRS can handle multicopy gene families since DLRS, apart from modeling sequence evolution, models gene duplication and loss using a gene evolution model (Arvestad L, Lagergren J, Sennblad B. 2009. The gene evolution model and computing its associated probabilities. J ACM. 56(2):1-44). We evaluate MixTreEM-DLRS using synthetic and biological data, and compare its performance with a recent genome-scale species tree reconstruction method PHYLDOG (Boussau B, Szöllősi GJ, Duret L, Gouy M, Tannier E, Daubin V. 2013. Genome-scale coestimation of species and gene trees. Genome Res. 23(2):323-330) as well as with a fast parsimony-based algorithm Duptree (Wehe A, Bansal MS, Burleigh JG, Eulenstein O. 2008. Duptree: a program for large-scale phylogenetic analyses using gene tree parsimony. Bioinformatics 24(13):1540-1541). Our method is competitive with PHYLDOG in terms of accuracy and runs significantly faster and our method outperforms Duptree in accuracy. The analysis constituted by MixTreEM without DLRS may also be used for selecting the target species tree, yielding a fast and yet accurate algorithm for larger data sets. MixTreEM is freely available at http://prime.scilifelab.se/mixtreem/.

Learning structures of Bayesian networks for variable groups

Abstract Bayesian networks, and especially their structures, are powerful tools for representing conditional independencies and dependencies between random variables. In applications where related variables form a priori known groups, chosen to represent different “views” to or aspects of the same entities, one may be more interested in modeling dependencies between groups of variables rather than between individual variables. Motivated by this, we study prospects of representing relationships between variable groups using Bayesian network structures. We show that for dependency structures between groups to be expressible exactly, the data have to satisfy the so-called groupwise faithfulness assumption. We also show that one cannot learn causal relations between groups using only groupwise conditional independencies, but also variable-wise relations are needed. Additionally, we present algorithms for finding the groupwise dependency structures.