Methods and systems for producing an expanded training set for machine learning using biological sequences

A computer-implemented system for training a supervised machine learning model with an expanded training set using biological sequences, comprising one or more computer processors and one or more storage devices having instructions stored thereon that, when executed by the one or more computer processors, cause the one or more computer processors to perform operations comprising:

• obtaining a set of original biological sequences comprising deoxyribonucleic acid (DNA) sequences, ribonucleic acid (RNA) sequences, or protein sequences;
• obtaining saliency values corresponding to one or more elements in each of the set of original biological sequences, wherein a saliency value in the saliency values corresponding to an element of the one or more elements indicates a degree of pertinence of the element to biological function, wherein the saliency values are derived from one or more of:
• evolutionary conservation across at least two different species, allele frequency in a human population of at least two humans, DNA accessibility, chromatin immunoprecipitation sequencing (ChIP-Seq), cross-linking immunoprecipitation sequencing (CLIP-Seq), systematic evolution of ligands by exponential enrichment (SELEX), a massively parallel reporter assay, and a mutational study;
• applying one or more modifications to each of the set of original biological sequences to create a set of modified biological sequences, and associating each of the set of modified biological sequences with an original biological sequence, wherein each of the set of original biological sequences has an associated label, and wherein each of the set of modified biological sequences is associated with the same label as the associated original biological sequence;
• generating one or more elements in each of the set of modified biological sequences using one or more elements in the associated original biological sequence and the corresponding saliency values, wherein a probability that an element in each of the set of modified biological sequences is the same as the elements in the associated original biological sequence is higher for larger corresponding saliency values, and wherein a biological function of the set of modified biological sequences is maintained relative to the associated original biological sequence;
• creating an expanded training set comprising the obtained set of original biological sequences and the set of modified biological sequences;
• and
• training the supervised machine learning model using the expanded training set.