Deep learning particle classification platform

A method implemented by a computing system of one or more processors, the method comprising:

• executing a software-platform associated with analyzing received events comprising single cell phenotype data, the received events being obtained via an acquisition device or based one or more datasets stored by a database, and the software-platform sorting cell populations indicated in the events;
• causing presentation, via the software-platform, of an interactive user interface, wherein the interactive user interface:
• receives user input indicating selection of a deep-learning module, the deep-learning module being obtained via a network to supplement the software-platform,
• wherein, based on the received events, the deep-learning module determines whether to select a machine learning model of a plurality of machine learning models accessible via the software-platform, wherein the selection is based on same similarities between phenotypes indicated in the events as phenotypes used for training the machine learning models, and wherein the phenotypes include, at least, surface markers,
• wherein upon a positive determination of selection of a machine learning model, the interactive user interface presents information identifying the selected machine learning model,
• and wherein upon a negative determination of selection of a machine learning model, the deep-learning module is configured to aggregate the events and train a new machine learning model,
• wherein the aggregated events include the received events and events associated with one or more users of the software platform, wherein the aggregated events are bulk normalized via performance of a transformation of unique molecular identifiers (UMI) counts and subsequently normalized to remove inter-experimental batch effects,
• wherein based on an accuracy associated with the new machine learning model exceeding a threshold, the deep-learning module is configured to select the new machine learning model and upload the new machine learning model to an online repository for access by users of the software platform;
• applying the selected machine learning model via generating a forward pass of information generated based on the events, the information being normalized based on the events;
• and
• presenting, via the interactive user interface, a graphical representation of cell populations indicated in the events, the graphical representation sorting the cell populations according to phenotypes, wherein the interactive user interface is updated to include the graphical representation in an analysis portion.

Non-transitory computer storage media storing instructions that when executed by a system of one or more processors, cause the one or more processors to perform operations comprising:

• executing a software-platform associated with analyzing events obtained via an acquisition device or based one or more datasets stored by a database, the software-platform sorting cell populations indicated in the events;
• causing presentation, via the software-platform, of an interactive user interface, wherein the interactive user interface:
• receives user input indicating selection of a deep-learning module, the deep-learning module being obtained via a network to supplement the software-platform,
• wherein, based on the received events, the deep-learning module determines whether to select a machine learning model of a plurality of machine learning models accessible via the software-platform, wherein the selection is based on same similarities between phenotypes indicated in the events as phenotypes used for training the machine learning models, and wherein the phenotypes include, at least, surface markers,
• wherein upon a positive determination of selection of a machine learning model, the interactive user interface presents information identifying the selected machine learning model,
• and wherein upon a negative determination of selection of a machine learning model, the deep-learning module is configured to aggregate the events and train a new machine learning model,
• wherein the aggregated events include the received events and events associated with one or more users of the software platform, wherein the aggregated events are bulk normalized via performance of a transformation of unique molecular identifiers (UMI) counts and subsequently normalized to remove inter-experimental batch effects,
• wherein based on an accuracy associated with the new machine learning model exceeding a threshold, the deep-learning module is configured to select the new machine learning model and upload the new machine learning model to an online repository for access by users of the software platform;
• applying the selected machine learning model via generating a forward pass of information generated based on the events, the information being normalized based on the events;
• and
• presenting, via the interactive user interface, information identifying assignments of respective classifications to the events, individual classifications reflecting individual cell types, and the assignments being based on output of the selected machine learning model, wherein the interactive user interface is updated to include the information in an analysis portion.