Summary/AbstractAlignment of biological sequences is a key step in understanding their evolution function and patterns ofactivity. Here we describe Machine Learning approaches to improve both accuracy and speed of highly-sensitive sequence alignment. To improve accuracy we develop methods to reduce erroneous annotationcaused by (1) the existence of low complexity and repetitive sequence and (2) the overextension ofalignments of true homologs into unrelated sequence. We describe approaches based on both hiddenMarkov models and Artificial Neural Networks to dramatically reduce these sorts of sequence annotationerror. We also address the issue of annotation speed with development of a custom Deep Learningarchitecture designed to very quickly filter away large portions of candidate sequence comparisons prior tothe relatively-slow sequence-alignment step. The results of these efforts will be incorporated into forks of theopen source sequence alignment tools HMMER MMSeqs and (where appropriate) BLAST; we will alsowork with community developers of annotation pipelines such as RepeatMasker and IMG/M to incorporatethese approaches. The development and incorporation into these widely used bioinformatics tools will leadto widespread impact on sequence annotation efforts.