Predicting likelihood and site of metastasis from patient records

1. A method for predicting metastasis of a cancer in a subject, comprising:

at a computer system having one or more processors, and memory storing one or more programs for execution by the one or more processors:

(A) obtaining, in electronic format, a plurality of at least 10,000 sequence reads, wherein the plurality of sequence reads is obtained for a plurality of RNA molecules from a sample of the cancer obtained from the subject

(B) determining from the plurality of at least 10,000 sequence reads, a plurality of data elements for the subject's cancer comprising:

a first set of sequence features comprising abundance values for the expression of a plurality of at least 30 genes in the sample of the cancer obtained from the subject; and

(C) applying, to the plurality of data elements for the subject's cancer comprising the first set of sequence features comprising abundance values for the expression of the plurality of at least 30 genes, one or more models that are collectively trained to provide a respective one or more indications of whether the cancer will metastasize in the subject,

thereby predicting whether the cancer will metastasize.

2. The method of claim 1 , wherein the plurality of data elements further comprises one or more personal characteristics about the subject selected from the group consisting of age, gender, and race and wherein the (C) applying includes applying the one or more personal characteristics about the subject to the one or more models.

3. The method of claim 1 , wherein the plurality of data elements further comprises one or more clinical features related to the diagnosis or treatment of the cancer in the subject selected from the group consisting of a stage of the cancer, a histopathological grade of the cancer, a therapy administered to the subject, a symptom associated with cancer or metastasis thereof, and a comorbidity with the cancer and wherein the (C) applying includes applying the one or more clinical features to the one or more models.

4. The method of claim 1 , wherein the plurality of data elements further comprises one or more temporal features related to the diagnosis or treatment of the cancer in the subject selected from the group consisting of a first temporal element indicating a duration of time since a diagnosis for the cancer, a second temporal element indicating a duration of time since an administration of a therapy to the subject, a third temporal element indicating a duration of time since an experience of a symptom associated with cancer or metastasis thereof, and a fourth temporal element indicating a duration of time since an experience of a comorbidity with the cancer and wherein the (C) applying includes applying the one or more temporal features to the one or more models.

5. The method of claim 1 , wherein the plurality of at least 30 genes comprises at least 20 genes selected from the group consisting of the genes listed in Table 2.

6. The method of claim 1 , wherein the plurality of at least 30 genes is no more than 250 genes.

7. The method of claim 1 , wherein the plurality of data elements further comprises a single-sample gene set enrichment analysis (ssGSEA) score for the transcriptional profile of the cancer and wherein the (C) applying includes applying the single-sample gene set enrichment analysis (ssGSEA) score to the one or more models.

8. The method of claim 1 , wherein the plurality of data elements further comprises a smoking status or a menopausal status of the subject and wherein the (C) applying includes applying the smoking status or the menopausal status to the one or more models.

9. The method of claim 1 , wherein the plurality of data elements further comprises a physical characteristic of the sample of the cancer and wherein the (C) applying includes applying the physical characteristic to the one or more models.

10. The method of claim 1 , wherein the plurality of data elements further comprises a mutational status for one or more genes in the sample of the cancer and wherein the (C) applying includes applying the mutational status for the one or more genes to the one or more models.

11. The method of claim 10 , wherein the mutational status is for a gene selected from the group consisting of the genes listed in Table 2.

12. The method of claim 1 , wherein the plurality of data elements further comprises a mutational status for one or more genes determined from genomic fragments of a non-cancerous tissue of the subject and wherein the (C) applying includes applying the mutational status for the one or more genes to the one or more models.

13. The method of claim 1 , wherein the plurality of data elements further comprises a copy number status for one or more genomic regions associated with cancer and wherein the (C) applying includes applying the copy number status for the one or more genomic regions to the one or more models.

14. The method of claim 1 , wherein the one or more models is a set of models that are collectively trained to provide, for each respective tissue in a plurality of tissues, a corresponding set of indications, in the one or more indications, of whether the cancer will metastasize to the respective tissue in the subject, wherein the corresponding set of indications includes a corresponding indication for each respective time horizon in a corresponding plurality of time horizons.

15. The method of claim 14 , wherein the set of models comprises, for each respective tissue in the plurality of tissues, a respective subset of models, wherein each respective model, in the respective subset of models, is trained to provide a respective indication of whether the cancer in the subject will metastasize to the respective tissue in the subject within a respective time horizon in the corresponding plurality of time horizons.

16. The method of claim 14 , wherein:

the corresponding plurality of time horizons for each respective tissue in the plurality of tissues is the same plurality of time horizons; and

the set of models comprises, for each respective time horizon in the plurality of time horizons, a respective model trained to provide, for each respective tissue in the plurality of tissues, a corresponding indication of whether the cancer in the subject will metastasize to the respective tissue in the subject within the respective time horizon.

17. The method of claim 14 , wherein the set of models comprises, for each respective tissue in the plurality of tissues, a corresponding model trained to provide, for each respective time horizon in the corresponding plurality of time horizons, a corresponding indication of whether the cancer in the subject will metastasize to the respective tissue in the subject within the respective time horizon.

18. The method of claim 14 , wherein:

the set of models comprises a respective model trained to provide, for each respective time horizon in a plurality of time horizons, a respective indication of whether the cancer will metastasize to any tissue in the subject; and

the plurality of indications of whether the cancer will metastasize includes, for each respective time horizon in the plurality of time horizons, a corresponding indication of whether the cancer will metastasize to any tissue in the subject.

19. The method of claim 14 , wherein the plurality of tissues comprises lymph tissue, liver tissue, and lung tissue.

20. The method of claim 14 , the method further comprising:

(D) generating a clinical report comprising the one or more indications of whether the cancer will metastasize; and

(E) displaying the clinical report in a graphical user interface (GUI), wherein the GUI comprises an anatomical representation of a body and a first affordance configured for switching between respective time horizons in the plurality of time horizons, the displaying comprising:

displaying a first rendering of metastatic predictions comprising, for each respective tissue in the plurality of tissues, a corresponding visual representation of the respective indication, in the plurality of indications, corresponding to whether the cancer in the subject will metastasize to the respective tissue within a first respective time horizon, in the corresponding plurality of time horizons, wherein the rendering is superposed upon the anatomical representation of the body; and

responsive to receiving a user input corresponding to the first affordance on the GUI, replacing display of the first rendering of metastatic predictions with display of a second rendering of metastatic predictions comprising, for each respective tissue in the plurality of tissues, a corresponding visual representation of the respective indication, in the plurality of indications, corresponding to whether the cancer in the subject will metastasize to the respective tissue within a second respective time horizon in the corresponding plurality of time horizons, wherein the rendering is superposed upon the anatomical representation of the body.

21. The method of claim 1 , wherein the one or more models comprises a trained survival function.

22. The method of claim 1 , the method further comprising:

when the one or more indications of whether the cancer will metastasize satisfies a first threshold risk for metastasis of the cancer, administering a first therapy tailored for treatment of metastatic cancer; and

when the one or more indications of whether the cancer will metastasize does not satisfy the first threshold risk for metastasis of the cancer, administering a second therapy tailored for treatment of non-metastatic cancer.

23. The method of claim 1 , the method further comprising, prior to the (C) applying, training the one or more models

using (i) values for the corresponding plurality of data elements across a plurality of training subjects that have cancer, wherein a portion of the plurality of training subjects have metastasized cancer and a portion of the training subjects have cancer that has not metastasized, and wherein the corresponding plurality of data elements serve as independent variables in the training and (ii) a corresponding indication for each respective training subject in the plurality of training subjects, of whether the respective training subject's cancer metastasized, wherein the indication serves as a dependent variable in the training

thereby obtaining the one or more models that are collectively trained to provide a respective one or more indications of whether the cancer will metastasize in a subject.

24. The method of claim 1 , the method further comprising generating a clinical report comprising the one or more indications of whether the cancer will metastasize.

25. The method of claim 1 , wherein the plurality of at least 10,000 sequence reads is at least 100,000 sequence reads.

26. The method of claim 1 , wherein the plurality of at least 10,000 sequence reads is at least 1,000,000 sequence reads.

27. The method of claim 1 , wherein the abundance values for the expression of a plurality of at least 30 genes are relative abundance values.

28. A computer system having one or more processors, and memory storing one or more programs for execution by the one or more processors, the one or more programs comprising instructions for performing a method for predicting metastasis of a cancer in a subject, the method comprising:

(A) obtaining, in electronic format, a plurality of at least 10,000 sequence reads, wherein the plurality of sequence reads is obtained for a plurality of RNA molecules from a sample of the cancer obtained from the subject

(B) determining from the plurality of at least 10,000 sequence reads, a plurality of data elements for the subject's cancer comprising:

a first set of sequence features comprising abundance values for the expression of a plurality of at least 30 genes in the sample of the cancer obtained from the subject and

(C) applying, to the plurality of data elements for the subject's cancer comprising the first set of sequence features comprising abundance values for the expression of the plurality of at least 30 genes, one or more models that are collectively trained to provide a respective one or more indications of whether the cancer will metastasize in the subject, thereby predicting whether the cancer will metastasize.

29. A non-transitory computer readable storage medium storing one or more programs configured for execution by a computer, the one or more programs comprising instructions for carrying out a method for predicting metastasis of a cancer in a subject, the method comprising:

(A) obtaining, in electronic format, a plurality of at least 10,000 sequence reads, wherein the plurality of sequence reads is obtained for a plurality of RNA molecules from a sample of the cancer obtained from the subject

(B) determining from the plurality of at least 10,000 sequence reads, a plurality of data elements for the subject's cancer comprising:

a first set of sequence features comprising abundance values for the expression of a plurality of at least 30 genes in the sample of the cancer obtained from the subject and

(C) applying, to the plurality of data elements for the subject's cancer comprising the first set of sequence features comprising abundance values for the expression of the plurality of at least 30 genes, one or more models that are collectively trained to provide a respective one or more indications of whether the cancer will metastasize in the subject, thereby predicting whether the cancer will metastasize.