A Simulation-Based Approach to Statistical Alignment

Abstract

Classic alignment algorithms utilize scoring functions which maximize similarity or minimize edit distances. These scoring functions account for both insertion-deletion (indel) and substitution events. In contrast, alignments based on stochastic models aim to explicitly describe the evolutionary dynamics of sequences by inferring relevant probabilistic parameters from input sequences. Despite advances in stochastic modeling during the last two decades, scoring-based methods are still dominant, partially due to slow running times of probabilistic approaches. Alignment inference using stochastic models involves estimating the probability of events, such as the insertion or deletion of a specific number of characters. In this work, we present SimBa-SAl, a simulation-based approach to statistical alignment inference, which relies on an explicit continuous Markov process for both indels and substitutions. SimBa-SAl has several advantages. First, using simulations, it decouples the estimation of event probabilities from the inference stage, which allows the introduction of accelerations to the alignment inference procedure. Second, it is general and can accommodate various stochastic models of indel formation. Finally, it allows computing the maximum-likelihood alignment, the probability of a given pair of sequences integrated over all possible alignments, and sampling alternative alignments according to their probability. We first show that SimBa-SAl allows accurate estimation of parameters of the long-indel model previously developed by Miklós, Lunter and Holmes in 2004. We next show that SimBa-SAl is more accurate than previously developed pairwise alignment algorithms, when analyzing simulated as well as empirical datasets. Finally, we study the goodness-of-fit of the long-indel and TKF91 models. We show that while the long-indel model fits the datasets better than TKF91, there is still room for improvement concerning the realistic modeling of evolutionary sequence dynamics.

A Model of Indel Evolution by Finite-State, Continuous-Time Machines

“…This illustrates a fundamental problem with the purely simulationbased LAHP19 approach; running it enough times to sample all cases is impractical. This may be one reason why the authors of LAHP19 limited the maximum gap length G to 50 residues (Levy Karin et al, 2019). A potential solution to this problem would be to use a limited sample to fit a parameteric model, although we have not explored that approach.…”

“…LG05 (Löytynoja and Goldman, 2005), RS07 (Redelings and Suchard, 2007), LAHP19 (Levy Karin et al, 2019), and DM20 (De Maio, 2020).…”

“…A purely simulation-based approach to estimating the gap probabilities of the GGI model has recently been described (Levy Karin et al, 2019). In the limit of an infinite number of random trials, this approach is exact.…”

A probabilistic model for indel evolution: differentiating insertions from deletions

“…Recently, advances have been made in the development of statistical alignment methods, allowing simultaneous model parameter inference and alignment (Suchard & Redelings, 2006;Novák et al, 2008;Bradley et al, 2009;Nute et al, 2019). Miklós et al (2004) and Levy-Karin et al (2019) have developed the longindel model, in which both indels and substitutions evolve along a phylogenetic tree assuming a joint continuous time Markov process. The rate of indels in this model depends on the indel length.…”

Statistical compression of protein sequences and inference of marginal probability landscapes over competing alignments using finite state models and Dirichlet priors

“…Beyond the problem with tuning alignment parameters, another major shortcoming is that most current programs report a single alignment. Many studies have demonstrated that optimizing an objective function under a specified model of sequence relatedness does not necessarily imply an optimization in homology (Fitch and Smith, 1983; Durbin et al , 1998; Levy Karin et al , 2019; Redelings and Suchard, 2005; Rosenberg, 2009) (see Supplementary notes for a more detailed discussion).…”