US 12,260,939 B2
Systems and methods for predicting compounds associated with transcriptional signatures
Juan Corchado Garcia, Cambridge, MA (US); Ragy Haddad, Cambridge, MA (US); and Diogo Camacho, Sudbury, MA (US)
Assigned to CELLARITY, INC., Somerville, MA (US)
Filed by Cellarity, Inc., Somerville, MA (US)
Filed on Dec. 13, 2023, as Appl. No. 18/539,190.
Claims priority of provisional application 63/387,251, filed on Dec. 13, 2022.
Prior Publication US 2024/0194299 A1, Jun. 13, 2024
Int. Cl. G06N 20/20 (2019.01); G16B 5/00 (2019.01); G16B 20/00 (2019.01); G16B 40/20 (2019.01)
CPC G16B 40/20 (2019.02) [G06N 20/20 (2019.01); G16B 5/00 (2019.02); G16B 20/00 (2019.02)] 32 Claims
OG exemplary drawing
 
1. A method of screening a plurality of test chemical compounds against a reference compound, the method comprising:
(A) obtaining a fingerprint of a chemical structure of each test chemical compound in the plurality of test chemical compounds, wherein the plurality of test chemical compounds comprises at least ten thousand test chemical compounds;
(B) obtaining, from one or more reference assay experiments, a plurality of abundance values for each cellular constituent in a set of cellular constituents across a first plurality of cells that have been exposed to a control solution free of the plurality of test chemical compounds;
(C) for each test chemical compound in the plurality of test chemical compounds, responsive to inputting the fingerprint of the chemical structure of the test chemical compound and the plurality of abundance values into a first model, retrieving, as output from the first model, a predicted similarity between (i) a predicted perturbational effect of the test chemical compound across the set of cellular constituents and (ii) a measured cell-based perturbational effect of the reference chemical compound across the set of cellular constituents, wherein, when the predicted similarity achieves a threshold similarity, the test chemical compound is associated with the reference compound thereby identifying a subset of the plurality of test chemical compounds in the plurality of test chemical compounds that are associated with the reference compound and wherein the first model comprises a first plurality of parameters;
(D) exposing a second plurality of cells to a test chemical compound in the subset of the plurality of test chemical compounds, wherein the test chemical compound has a Tanimoto coefficient of less than 0.85 with respect to the reference chemical compound;
(E) measuring a transcriptional response of each gene in a panel of genes in the second plurality of cells after the exposing (D);
(F) exposing a third plurality of cells to the reference compound;
(G) measuring a transcriptional response of each gene in the panel of genes in the third plurality of cells after the exposing (F); and
(H) determining that a transcriptional response to the test compound and the reference compound is similar across the panel of genes thereby validating the perturbational effect of the test chemical compound.