Calculating cell-type RNA profiles for diagnosis and treatment

A computer system for training a model for determining a cancer composition of a subject, the computer system comprising:

• at least one processor;
• and
• a memory, the memory storing at least one program for execution by the at least one processor, the at least one program comprising instructions for:
• obtaining, in electronic form, a set of relative RNA abundance values comprising, for each respective training subject in a plurality of training subjects, for each respective genetic target in a plurality of genetic targets, a corresponding relative RNA abundance value for the respective genetic target from bulk-cell RNA sequencing of one or more corresponding biological samples obtained from a corresponding tumor specimen of the respective training subject, wherein the plurality of training subjects is more than 100 training subjects and wherein no single-cell RNA-sequencing data or cell-type-specific expression profiles are provided to the computer system;
• obtaining, in electronic form, a set of observed expression distribution parameters comprising, for each respective genetic target in the plurality of genetic targets, a corresponding observed shape parameter and a corresponding observed mean parameter by fitting a corresponding statistical distribution to a corresponding subset of the set of relative RNA abundance values comprising, for each respective training subject in the plurality of training subjects, the corresponding relative mRNA abundance value for the respective genetic target;
• obtaining, in electronic form, a set of cell proportions comprising, for each respective training subject in the plurality of training subjects, for each respective cell type in a plurality of cell types, a corresponding proportion of the respective cell type in the one or more corresponding biological samples;
• refining a first optimization model to correlate RNA abundance values for the plurality of genetic targets with proportions of the plurality of cell types, subject to a first plurality of constraints, the first plurality of constraints including:
• (i) the corresponding observed shape parameter for each respective genetic target in the plurality of genetic targets,
• (ii) the set of cell proportions, and
• (iii) the set of relative RNA abundance values, thereby generating, for each respective cell type in the plurality of cell types, a corresponding cell type RNA expression profile.

A non-transitory computer-readable storage medium having stored thereon program code instructions that, when executed by a processor, cause the processor to perform a method for determining a cancer composition of a subject, the method comprising:

• obtaining, in electronic form, a set of relative RNA abundance values comprising, for each respective training subject in a plurality of training subjects, for each respective genetic target in a plurality of genetic targets, a corresponding relative RNA abundance value for the respective genetic target from bulk-cell RNA sequencing of one or more corresponding biological samples obtained from a corresponding tumor specimen of the respective training subject, wherein the plurality of training subjects is more than 100 training subjects and wherein no single-cell RNA-sequencing data or cell-type-specific expression profiles are provided to the computer system;
• obtaining, in electronic form, a set of observed expression distribution parameters comprising, for each respective genetic target in the plurality of genetic targets, a corresponding observed shape parameter and a corresponding observed mean parameter by fitting a corresponding statistical distribution to a corresponding subset of the set of relative RNA abundance values comprising, for each respective training subject in the plurality of training subjects, the corresponding relative mRNA abundance value for the respective genetic target;
• obtaining, in electronic form, a set of cell proportions comprising, for each respective training subject in the plurality of training subjects, for each respective cell type in a plurality of cell types, a corresponding proportion of the respective cell type in the one or more corresponding biological samples;
• refining a first optimization model to correlate RNA abundance values for the plurality of genetic targets with proportions of the plurality of cell types, subject to a first plurality of constraints, the first plurality of constraints including:
• (i) the corresponding observed shape parameter for each respective genetic target in the plurality of genetic targets,
• (ii) the set of cell proportions, and
• (iii) the set of relative RNA abundance values, thereby generating, for each respective cell type in the plurality of cell types, a corresponding cell type RNA expression profile.