MOTIVATION: Deep-learning models, such as convolutional neural networks, are able to accurately map biological sequences to associated functional readouts and properties by learning predictive de novo representations. In silico saturation mutagenesis (ISM) is a popular feature attribution technique for inferring contributions of all characters in an input sequence to the model's predicted output. The main drawback of ISM is its runtime, as it involves multiple forward propagations of all possible mutations of each character in the input sequence through the trained model to predict the effects on the output. RESULTS: We present fastISM, an algorithm that speeds up ISM by a factor of over 10× for commonly used convolutional neural network architectures. fastISM is based on the observations that the majority of computation in ISM is spent in convolutional layers, and a single mutation only disrupts a limited region of intermediate layers, rendering most computation redundant. fastISM reduces the gap between backpropagation-based feature attribution methods and ISM. It far surpasses the runtime of backpropagation-based methods on multi-output architectures, making it feasible to run ISM on a large number of sequences. AVAILABILITY AND IMPLEMENTATION: An easy-to-use Keras/TensorFlow 2 implementation of fastISM is available at https://github.com/kundajelab/fastISM. fastISM can be installed using pip install fastism. A hands-on tutorial can be found at https://colab.research.google.com/github/kundajelab/fastISM/blob/master/notebooks/colab/DeepSEA.ipynb. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: In silico saturation mutagenesis (ISM) is a popular approach in computational genomics for calculating feature attributions on biological sequences that proceeds by systematically perturbing each position in a sequence and recording the difference in model output. However, this method can be slow because systematically perturbing each position requires performing a number of forward passes proportional to the length of the sequence being examined. RESULTS: In this work, we propose a modification of ISM that leverages the principles of compressed sensing to require only a constant number of forward passes, regardless of sequence length, when applied to models that contain operations with a limited receptive field, such as convolutions. Our method, named Yuzu, can reduce the time that ISM spends in convolution operations by several orders of magnitude and, consequently, Yuzu can speed up ISM on several commonly used architectures in genomics by over an order of magnitude. Notably, we found that Yuzu provides speedups that increase with the complexity of the convolution operation and the length of the sequence being analyzed, suggesting that Yuzu provides large benefits in realistic settings. AVAILABILITY AND IMPLEMENTATION: We have made this tool available at https://github.com/kundajelab/yuzu.

MOTIVATION: Successful science often involves not only performing experiments well, but also choosing well among many possible experiments. In a hypothesis generation setting, choosing an experiment well means choosing an experiment whose results are interesting or novel. In this work, we formalize this selection procedure in the context of genomics and epigenomics data generation. Specifically, we consider the task faced by a scientific consortium such as the National Institutes of Health ENCODE Consortium, whose goal is to characterize all of the functional elements in the human genome. Given a list of possible cell types or tissue types ('biosamples') and a list of possible high-throughput sequencing assays, where at least one experiment has been performed in each biosample and for each assay, we ask 'Which experiments should ENCODE perform next?' RESULTS: We demonstrate how to represent this task as a submodular optimization problem, where the goal is to choose a panel of experiments that maximize the facility location function. A key aspect of our approach is that we use imputed data, rather than experimental data, to directly answer the posed question. We find that, across several evaluations, our method chooses a panel of experiments that span a diversity of biochemical activity. Finally, we propose two modifications of the facility location function, including a novel submodular-supermodular function, that allow incorporation of domain knowledge or constraints into the optimization procedure. AVAILABILITY AND IMPLEMENTATION: Our method is available as a Python package at https://github.com/jmschrei/kiwano and can be installed using the command pip install kiwano. The source code used here and the similarity matrix can be found at http://doi.org/10.5281/zenodo.3708538. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

We present apricot, an open source Python package for selecting representative subsets from large data sets using submodular optimization. The package implements several efficient greedy selection algorithms that offer strong theoretical guarantees on the quality of the selected set. Additionally, several submodular set functions are implemented, including facility location, which is broadly applicable but requires memory quadratic in the number of examples in the data set, and a feature-based function that is less broadly applicable but can scale to millions of examples. Apricot is extremely efficient, using both algorithmic speedups such as the lazy greedy algorithm and memoization as well as code optimization using numba. We demonstrate the use of subset selection by training machine learning models to comparable accuracy using either the full data set or a representative subset thereof. This paper presents an explanation of submodular selection, an overview of the features in apricot, and applications to two data sets.