Systems and methods for detecting biological features
A computer having a memory stores instructions for receiving data. The data comprises one or more characteristics for each cellular constituent in a plurality of cellular constituents that have been measured in a test organism of a species or a test biological specimen from an organism of the species. The memory further stores instructions for computing a model in a plurality of models, wherein the model is characterized by a model score that represents the likelihood of a biological feature in the test organism or the test biological specimen. Computation of the model comprises determining the model score using one or more characteristics for one or more cellular constituents in the plurality of cellular constituents. The memory also stores instructions for repeating the instructions for computing one or more times, thereby computing the plurality of models. The memory also stores instructions for communicating computed model scores.
1. A method comprising:
receiving data, wherein said data comprises one or more characteristics for each cellular constituent in a plurality of cellular constituents that have been measured in a single biological specimen from a patient;
applying a plurality of pre-trained models using said data to produce a model score for each respective pre-trained model in said plurality of pre-trained models thereby obtaining a plurality of model scores, wherein each model score in the plurality of model scores is a number that represents the likelihood of a corresponding biological feature represented by the corresponding model being present in the patient, wherein prior to applying the plurality of pre-trained models it is deemed unknown whether the patient has the corresponding biological feature, wherein applying a respective model in said plurality of models comprises determining the model score for the respective model using one or more characteristics for a sub-plurality of cellular constituents in said plurality of cellular constituents in said data that have been measured in the patient, the plurality of pre-trained models comprising a first pre-trained model using one or more characteristics of a first sub-plurality of cellular constituents in said plurality of cellular constituents and a second pre-trained model using one or more characteristics of a second sub-plurality of cellular constituents in said plurality of cellular constituents, wherein the first sub-plurality of cellular constituents includes at least one cellular constituent that is not present in the second sub-plurality of cellular constituents;
communicating said plurality of model scores; and
treating the patient for a biological feature associated with a model score representing that there is a high likelihood the biological feature is present in the patient.
2. The method of claim 1 , wherein:
said one or more characteristics are abundances;
said cellular constituents are mRNAs;
said applying said plurality of pre-trained models comprises determining said model score for each respective model using said abundances of more than 500 mRNAs in said plurality of cellular constituents that have been measured in said biological specimen;
said biological features comprise biological features that are different diseases; and
said different diseases are each a primary tumor site of origin.
3. The method of claim 2 , wherein said more than 500 mRNAs are more than 1000 mRNAs.
4. The method of claim 2 , wherein said primary tumor sites of origin are selected from the group consisting of prostate, breast, colorectum, lung, liver, gastroesophagus, pancreas, ovary, kidney, and bladder/ureter.
5. The method of claim 2 , wherein the patient is human.
6. The method of claim 1 , wherein said plurality of pre-trained models collectively represent the independent likelihoods of two or more biological features wherein at least one of said biological features is a disease and at least one of said biological features is sensitivity to a drug.
7. The method of claim 1 , wherein
a model precondition is included within a first pre-trained model in the plurality of pre-trained models, and
applying the plurality of pre-trained models comprises determining whether the model precondition has been satisfied, wherein satisfaction of the model precondition for the first pre-trained model requires a second pre-trained model in the plurality of pre-trained models to have a specific predetermined model score, and wherein the first pre-trained model is applied when the model precondition is satisfied and the first pre-trained model is not applied when the model precondition is not satisfied.
8. The method of claim 1 , wherein said biological feature is a disease.
9. The method of claim 1 , wherein the patient is human.
10. The method of claim 1 , wherein said one or more characteristics comprises cellular constituent abundance and said data comprises cellular constituent abundances of at least one hundred cellular constituents in the biological specimen from the patient.
11. The method of claim 1 , wherein said one or more characteristics comprises cellular constituent abundance and said data comprises cellular constituent abundances of at least five hundred cellular constituents in the biological specimen from the patient.
12. The method of claim 1 , wherein said one or more characteristics comprises cellular constituent abundance and said data comprises cellular constituent abundances of between one thousand and twenty thousand cellular constituents in the biological specimen from the patient.
13. The method of claim 1 , wherein said biological feature is sensitivity to a drug.
14. The method of claim 1 , wherein the plurality of pre-trained models collectively represent the likelihood of each of five or more biological features.
15. The method of claim 1 , wherein each biological feature in said five or more biological features is a cancer origin.
16. The method of claim 1 , wherein the steps of receiving data, applying a plurality of pre-trained models, and communicating said plurality of model scores are performed at a system comprising one or more processors and memory.
17. The method of claim 1 , wherein treating the patient comprises administering a drug to the patient.
18. The method of claim 1 , wherein treating the patient comprises administering a combination of drugs to the patient.