EP-4736128-A1 - SYSTEMS AND METHODS OF CELL SORTING IMPLEMENTING ARTIFICIAL INTELLIGENCE

Abstract

Disclosed herein are apparatuses, systems, as well as related methods, computing devices, and computer-readable media related to real time cell sorter cell sorting using embeddings. For example, in some embodiments a method may comprise receiving first cell sorter data. The first cell sorter data may include cell sorter data including microscopy data, hyperspectral imaging data, high-dimensional vector data, or one or more combinations thereof. In some embodiments, the cell sorter data may include quantitative fluorescence data expressed as one or more of antibodies bound per cell, antibody binding capacity (ABC), molecules of equivalent soluble fluorochrome (MESF), one or more other quantitative indicators of fluorescence, or one or more combinations thereof. In some embodiments, the quantitative fluorescence data includes one or more fluorescence signals from: one or more fluorescent proteins, one or more fluorescent dyes, one or more fluorescently conjugate antibodies, or one or more combinations thereof.

Inventors

• RUDY, Samuel Henry Adler
• FOX, Daniel Nelson
• ROHRBACKER, Nicholas John

Assignees

• Life Technologies Corporation

Dates

Publication Date

20260506

Application Date

20240627

Claims (20)

1. 1. A sy stem comprising : a cell sorting device, and a computing device configured to perform the steps of: collecting first cell sorter data associated with a first portion of a sample of one or more cells; training, based on one or more of the first cell sorter data, user input, a first representation of the first cell sorter data or a combination thereof, a machine learning model; receiving one or more gating instructions from a user input based on the first cell sorter data, the machine learning model, or a combination thereof; receiving second cell sorter data associated with a second portion of the sample of one or more cells; determining, based on the trained machine learning model and the user-inputted gating instructions, a classification of the second cell sorter data, a representation of the second cell sorter data, or a combination thereof; and sorting, based on the classification, a portion of the sample.
2. 2. The system of claim 1, wherein the first representation of the first cell sorter data is determined based on applying a mapping process to the first cell sorter data.
3. 3. A method of sorting a sample of particles, the method comprising: collecting first cell sorter data associated with a first portion of a sample of one or more particles comprising cells; training, based on one or more of the first cell sorter data, user input, a first representation of die first cell sorter data or a combination thereof, a machine learning model; receiving one or more gating instructions from a user input based on the first cell sorter data, the machine learning model, or a combination thereof; receiving second cell sorter data associated with a second portion of the sample of one or more cells; determining, based on the trained machine learning model and the user-inputted gating instructions, a classification of the second cell sorter data, a representation of the second cell sorter data, or a combination thereof; and sorting, based on the classification, a portion of the sample.
4. 4. The method of claim 3, wherein the one or more gating instructions are determined based on one or more of data output from the machine learning model, an embedding output from the machine learning model, a parametric embedding of the cell sorter data generated using the machine learning model, or a combination thereof.
5. 5. The method of claim 3 or 4, wherein the user input comprises gating instructions, sorting instructions, or a combination thereof indicating one or more groupings associated with the first cell sorter data.
6. 6. The method of claim 5, wherein the one or more groupings comprise one or more signal clusterings or gatings for one or more parameters comprising: removal of doublets, cell viability, light scatter, expression of one or more specific lineage markers, or one or more combinations thereof.
7. 7. The method of any one of claims 3 to 6. w herein the cell sorter comprises a field- programmable gate array (FPGA), wherein the FPGA is programmed with parameters of the trained machine learning model and configured to analyze a plurality of portions of the sample using the trained machine learning model, and configured to cause the FPGA to store or comprise one or more trained weights of the trained machine learning model.
8. 8. The method of any one of claims 3 to 7, w herein the first representation of the first cell sorter data is determined based on applying a mapping process to the first cell sorter data, the mapping process comprising one or more of a clustering process, a dimensionality reduction process, an embedding, a non-parametric embedding, or a combination thereof.
9. 9. The method of claim 8, w herein the embedding comprises a uniform Manifold Approximation and Projection (UMAP), a t-distributed Stochastic Neighbor Embedding (t-SNE), another nonlinear embedding, or a combination thereof.
10. 10. The method of any one of claims 3 to 9, wherein the machine learning model transforms cell sorter data having a higher number of dimensions to a representation of the transformed cell sorter data having a lower number of dimensions.
11. 11. The method of claim 10, wherein the transformed cell sorter data comprises a low dimensional representation optimized to preserve one or more features of a high dimensional representation.
12. 12. The method of claim 10, wherein the representation of the cell sorter data comprises one or more clusters of cell sorter data for separating into subsequent representations of data.
13. 13. The method of any one of claims 3 to 12, wherein the cell sorter data comprises quantitative fluorescence data expressed as one or more of antibodies bound per cell, antibody binding capacity (ABC), or molecules of equivalent soluble fluorochrome (MESF), one or more other quantitative indicators of fluorescence, or one or more combinations thereof.
14. 14. The method of claim 13, wherein the quantitative fluorescence data comprises one or more fluorescence signals.
15. 15. The method of claim 13, wherein the fluorescence signals are derived from one or more fluorescent proteins, one or more fluorescent dyes, one or more fluorescently conjugate antibodies, one or more populations of fluorescent beads or fluorescently labeled beads, or one or more combinations thereof.
16. 16. The method of any one of claims 3 to 15, wherein the machine learning model comprises a neural network.
17. 17. The method of claim 3, wherein the machine learning model is trained without prior determination of a first representation of the first cell sorter data.
18. 18. The method of claim 1 , wherein the neural network is trained to learn the mechanism of a mapping process for generating one or more representations of the cell sorter data for a cell sorter measurement session.
19. 19. The method of claim 16, wherein the neural network comprises one or more of an artificial neural network, a convolutional neural network, a recurrent neural netw ork, or one or more combinations thereof.
20. 20. The method of any one of claims 16, 18 or 19, w herein the neural network comprises one or more nonlinear activation functions.

Description

Systems and Methods of Cell Sorting Implementing Artificial Intelligence Cross-Reference To Related Applications [1] This Application claims the benefit of United States Provisional Application No. 63/510,959, filed June 29, 2023, the entirety of which is incorporated by reference herein. Background [2] Cell sorting generally involves sorting of cells from a heterogeneous sample of cells, but sorting cells presents many challenges. Current strategies for cell sorting based on nonlinear embeddings rely on algorithms that construct sequences of exclusionary gates drawn on two- dimensional density plots based on cell marker expression levels. These gates and sequences are subjective, potentially inexact, and incapable of using relations between three or more cell markers on a single gate. Thus, there is a need for more sophisticated techniques for sorting cells and related analysis. The present invention addresses this need. Brief Description of the Drawings [3] Embodiments will be readily understood by the following detailed description in conjunction with the accompanying drawings. To facilitate this description, like reference numerals designate like structural elements. Embodiments are illustrated by way of example, not by w ay of limitation, in the figures of the accompanying draw ings. The drawings include a number of features that are not discussed in detail herein for clarity7 of exposition, but the purpose and operation of these features w ill be understood by one of ordinary' skill in the art and may take any suitable form. Any of the features of the drawings may be used in combination with any suitable ones of the features of other of the accompanying drawings and/or in combination with any suitable ones of the features of the embodiments disclosed herein. [4] FIGS. 1A, IB and 1C are block diagrams of exemplary' cell sorter support modules for performing support operations, in accordance with various embodiments. [5] FIGS. 2A, 2B and 2C are flow diagrams of exemplary methods of performing support operations, in accordance w ith various embodiments. [6] FIG. 3 is an example of a graphical user interface that may be used in the performance of some or all of the support methods disclosed herein, in accordance with various embodiments. [7] FIG. 4 is a block diagram of an example computing device that may perform some or all of the cell sorter support methods disclosed herein, in accordance with various embodiments. [8] FIG. 5 is a block diagram of an example cell sorter support system in which some or all of the cell sorter support methods disclosed herein may be perfonned, in accordance w ith various embodiments. [9] FIGS. 6 A, 6B and 6C are exemplar}' workflows as contemplated by embodiments of the present disclosure. In FIG. 6A, the left panel depicts a spectral cell sorter recording cells as events. The spectral cell sorter is equipped to capture a detailed spectral signature of each cell. The middle panel is a graphical representation of a neural network for processing the spectral signature. The right panel depicts exemplary gating on a parametric embedding. FIG. 6B depicts an exemplary workflow using a neural network that is trained to reproduce a nonparametric embedding. FIG. 6C depicts an exemplary workflow using neural network that is trained to directly produce a parametric embedding. [10] FIGS. 7A. 7B and 7C depict a simple representative gating progression (progressing from 7A to 7B to 7C) of a cell sample sorted for CD4+ T cells using a spectral cell sorter. [11] FIGS. 8A and 8B depict plots of CD4+ T cells highlighted in both a tSNE (FIG. 8A) embedding and UMAP (FIG. 8B) embedding. [12] FIGS. 9A to 9C depict an exemplary selection of clusters from an embedding (FIG. 9A). and subsequent traditional gating sequences (FIGS. 9B and 9C) determined using a gate-finding algorithm as contemplated herein. [13] FIGS. 1 OA to 1 OD depict graphical representations of exemplary mapping processes as contemplated by the present disclosure. FIGS. 10A and 10B show the workflow using a neural network to reproduce a non-parametric embedding. FIGS. IOC and 10D shows the workflow using a network trained to produce the embedding directly. The top panels of FIG. 10A depict identical gates shown on an initial non -parametric UMAP embedding, the neural network scoring on the training data comprising 9901 events, and the neural network scoring of the remaining 255,975 events. The top panels of FIG. 10C depict gates shown on neural netw ork scoring of the training data comprising 9901 events, and the neural network scoring of the remaining 255,975 events. The lower panels show' box-and-whiskcr plots showing similarity of medians and percentile ranges for the above gates across embeddings. Detailed Description [14] Disclosed herein are apparatuses, systems, as well as related methods, computing devices, and computer-readable media related to real time cell sorter cell sorting using embeddings. For example, in some