Open Access

Background The alignment of large numbers of protein sequences is a challenging task and its importance grows rapidly along with the size of biological datasets. State-of-the-art algorithms have a tendency to produce less accurate alignments with an increasing number of sequences. This is a fundamental problem since many downstream tasks rely on accurate alignments. Results We present learnMSA, a novel statistical learning approach of profile hidden Markov models (pHMMs) based on batch gradient descent. Fundamentally different from popular aligners, we fit a custom recurrent neural network architecture for (p)HMMs to potentially millions of sequences with respect to a maximum a posteriori objective and decode an alignment. We rely on automatic differentiation of the log-likelihood, and thus, our approach is different from existing HMM training algorithms like Baum–Welch. Our method does not involve progressive, regressive, or divide-and-conquer heuristics. We use uniform batch sampling to adapt to large datasets in linear time without the requirement of a tree. When tested on ultra-large protein families with up to 3.5 million sequences, learnMSA is both more accurate and faster than state-of-the-art tools. On the established benchmarks HomFam and BaliFam with smaller sequence sets, it matches state-of-the-art performance. All experiments were done on a standard workstation with a GPU. Conclusions Our results show that learnMSA does not share the counterintuitive drawback of many popular heuristic aligners, which can substantially lose accuracy when many additional homologs are input. LearnMSA is a future-proof framework for large alignments with many opportunities for further improvements.

Convolutional Neural Network-based image classification models are the current state-of-the-art for solving image classification problems. However, obtaining and using such a model to solve a specific image classification problem presents several challenges in practice. To train the model, we need to find good hyperparameter values for training, such as initial model weights or learning rate. However, finding these values is usually a non-trivial process. Another problem is that the training data used for model training is often class-imbalanced in practice. This usually has a negative impact on model training. However, not only is it challenging to obtain a Convolutional Neural Network-based model, but also to use the model after model training. After training, the model might be applied to images that were drawn from a data distribution that is different from the data distribution the training data was drawn from. These images are typically referred to as out-of-distribution samples. Unfortunately, Convolutional Neural Network-based image classification models typically fail to predict the correct class for out-of-distribution samples without warning, which is problematic when such a model is used for safety-critical applications. In my work, I examined whether information from the layers of a Convolutional Neural Network-based image classification model (pixels and activations) can be used to address all of these issues. As a result, I suggest a method for initializing the model weights based on image patches, a method for balancing a class-imbalanced dataset based on layer activations, and a method for detecting out-of-distribution samples, which is also based on layer activations. To test the proposed methods, I conducted extensive experiments using different datasets. My experiments showed that layer information (pixels and activations) can indeed be used to address all of the aforementioned challenges when training and using Convolutional Neural Network-based image classification models.

We consider Walsh’s conformal map from the exterior of a compact set E ⊆ C onto a lemniscatic domain. If E is simply connected, the lemniscatic domain is the exterior of a circle, while if E has several components, the lemniscatic domain is the exterior of a generalized lemniscate and is determined by the logarithmic capacity of E and by the exponents and centers of the generalized lemniscate. For general E, we characterize the exponents in terms of the Green’s function of Ec. Under additional symmetry conditions on E, we also locate the centers of the lemniscatic domain. For polynomial pre-images E = P−1(Ω) of a simply-connected infinite compact set Ω, we explicitly determine the exponents in the lemniscatic domain and derive a set of equations to determine the centers of the lemniscatic domain. Finally, we present several examples where we explicitly obtain the exponents and centers of the lemniscatic domain, as well as the conformal map.

Background An important initial phase of arguably most homology search and alignment methods such as required for genome alignments is seed finding. The seed finding step is crucial to curb the runtime as potential alignments are restricted to and anchored at the sequence position pairs that constitute the seed. To identify seeds, it is good practice to use sets of spaced seed patterns, a method that locally compares two sequences and requires exact matches at certain positions only. Results We introduce a new method for filtering alignment seeds that we call geometric hashing. Geometric hashing achieves a high specificity by combining non-local information from different seeds using a simple hash function that only requires a constant and small amount of additional time per spaced seed. Geometric hashing was tested on the task of finding homologous positions in the coding regions of human and mouse genome sequences. Thereby, the number of false positives was decreased about million-fold over sets of spaced seeds while maintaining a very high sensitivity. Conclusions An additional geometric hashing filtering phase could improve the run-time, accuracy or both of programs for various homology-search-and-align tasks.