Systems and methods for diagnosing a biological specimen using probabilities
First Claim
1. A computer implemented method of determining, for each respective phenotypic characterization in a set of {T1, . . . , Tk} phenotypic characterizations, a probability that a test biological specimen has the respective phenotypic characterization, the method comprising:
- (A) learning a pairwise probability function gpq(X, Wpq) using a training population, for a pair of phenotypic characterizations (Tp, Tq) in the set of {T1, . . . , Tk} phenotypic characterizations, wherein(i) there are at least five training samples in the training population for each phenotypic characterization in the set of {T1, . . . , Tk} phenotypic characterizations;
(ii) Y is the set of all training samples in the training population that exhibits either phenotypic characterization Tp or phenotypic characterization Tq, and each Yi in Y is the set of {yi1, . . . , yin} cellular constituent abundance values for a plurality of cellular constituents measured from a sample i, from the training population, which exhibits either phenotypic characterization Tp or phenotypic characterization Tq;
(iii) Wpq is a set of parameters derived from Y in the learning step (A) for a pair of phenotypic characterizations (Tp, Tq) by substituting each Yi into gpq(X, Wpq), as X, during said learning step (A);
(iv) k is 3 or greater;
(v) n is at least 1; and
(vi) p is not equal to q;
(B) repeating the learning step (A) for a different pair of phenotypic characterizations (Tp, Tq), using the training population, for all unique pairs of phenotypic characterizations in the set of {T1 . . . , Tk} phenotypic characterizations, thereby deriving a plurality of pairwise probability functions G={g1,2(X, W1,2), . . . , gk-1, k(X, Wk-1, k)};
(C) computing a plurality of pairwise probability values P={p1,2, . . . , pk-1, k}, wherein each pairwise probability value ppq in P is equal to gpq(Z, Wpq) in G, the probability that the test biological specimen has phenotypic characterization Tp and does not have phenotypic characterization Tq, wherein Z is a set of {z1, . . . , zn} cellular constituent abundance values measured from the test biological specimen for said plurality of cellular constituents;
(D) optionally converting P to a set M of k probabilities, wherein M={p1, p2, . . . , pk}, wherein each probability pj in M is a probability for a phenotypic characterization in the set of {T1, . . . , Tk} phenotypic characterizations that the test biological specimen has the phenotypic characterization such that
8 Assignments
0 Petitions
Accused Products
Abstract
Apparatus, systems and methods for determining, for each respective phenotypic characterization in a set of {T1, . . . , Tk} characterizations, that a test specimen has the respective characterization are provided. A pairwise probability function gpq(X, Wpq), for a phenotypic pair (Tp, Tq) in {T1, . . . , Tk} is learned using a training population. Wpq is a set of parameters derived from Y for (Tp, Tq) by substituting each Y1 in Y into gpq(X, Wpq), as X, where Yi is the set of cellular constituent abundance values from sample i in the training population exhibiting Tp or Tq. The learning step is repeated for each (Tp, Tq) in {T1 . . . , Tk}, thereby deriving pairwise probability functions G={g1,2(X, W1,2), . . . , gk-1, k(X, Wk-1, k)}. Pairwise probability values P={p1,2, . . . , pk-1, k} are computed, where each ppq is equal to gpq(Z, Wpq) in G, the probability that the test specimen has Tp and not Tq, where Z is cellular constituent abundance values of the test specimen.
41 Citations
25 Claims
-
1. A computer implemented method of determining, for each respective phenotypic characterization in a set of {T1, . . . , Tk} phenotypic characterizations, a probability that a test biological specimen has the respective phenotypic characterization, the method comprising:
-
(A) learning a pairwise probability function gpq(X, Wpq) using a training population, for a pair of phenotypic characterizations (Tp, Tq) in the set of {T1, . . . , Tk} phenotypic characterizations, wherein (i) there are at least five training samples in the training population for each phenotypic characterization in the set of {T1, . . . , Tk} phenotypic characterizations; (ii) Y is the set of all training samples in the training population that exhibits either phenotypic characterization Tp or phenotypic characterization Tq, and each Yi in Y is the set of {yi1, . . . , yin} cellular constituent abundance values for a plurality of cellular constituents measured from a sample i, from the training population, which exhibits either phenotypic characterization Tp or phenotypic characterization Tq; (iii) Wpq is a set of parameters derived from Y in the learning step (A) for a pair of phenotypic characterizations (Tp, Tq) by substituting each Yi into gpq(X, Wpq), as X, during said learning step (A); (iv) k is 3 or greater; (v) n is at least 1; and (vi) p is not equal to q; (B) repeating the learning step (A) for a different pair of phenotypic characterizations (Tp, Tq), using the training population, for all unique pairs of phenotypic characterizations in the set of {T1 . . . , Tk} phenotypic characterizations, thereby deriving a plurality of pairwise probability functions G={g1,2(X, W1,2), . . . , gk-1, k(X, Wk-1, k)}; (C) computing a plurality of pairwise probability values P={p1,2, . . . , pk-1, k}, wherein each pairwise probability value ppq in P is equal to gpq(Z, Wpq) in G, the probability that the test biological specimen has phenotypic characterization Tp and does not have phenotypic characterization Tq, wherein Z is a set of {z1, . . . , zn} cellular constituent abundance values measured from the test biological specimen for said plurality of cellular constituents; (D) optionally converting P to a set M of k probabilities, wherein M={p1, p2, . . . , pk}, wherein each probability pj in M is a probability for a phenotypic characterization in the set of {T1, . . . , Tk} phenotypic characterizations that the test biological specimen has the phenotypic characterization such that - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20)
-
-
21. An apparatus for determining, for each respective phenotypic characterization in a set of {T1, . . . , Tk} phenotypic characterizations, a probability that a test biological specimen has the respective characterization, the apparatus comprising:
-
a processor; and a memory, coupled to the processor, the memory storing a module comprising; (A) instructions for learning a pairwise probability function gpq(X, Wpq) using a training population, for a pair of phenotypic characterizations (Tp, Tq) in the set of {T1, . . . , Tk} phenotypic characterizations, wherein; (i) there are at least five training samples in the training population for each phenotypic characterization in the set of {T1, . . . , Tk} phenotypic characterizations; (ii) Y is the set of all training samples in the training population that exhibits either phenotypic characterization Tp or phenotypic characterization Tq, and each Yi in Y is the set of {yi1, . . . , yin} cellular constituent abundance values for a plurality of cellular constituents measured from a sample i, from the training population, which exhibits either phenotypic characterization Tp or phenotypic characterization Tq; (iii) Wpq is a set of parameters derived from Y by the instructions for learning (A) for a pair of phenotypic characterizations (Tp, Tq) by substituting each Yi into gpq(X, Wpq), as X, during said learning step (A); (iv) k is 3 or greater; (v) n is at least 1; and (vi) p is not equal to q, (B) instructions for repeating the instructions for learning (A) for a different pair of phenotypic characterizations (Tp, Tq), using the training population, for all unique pairs of phenotypic characterizations in the set of {T1 . . . , Tk} phenotypic characterizations, thereby deriving a plurality of pairwise probability functions G={g1,2(X, W1,2), . . . , gk-1, k(X, Wk-1, k)}; (C) instructions for computing a plurality of pairwise probability values P={p1,2, . . . , pk-1, k}, wherein each pairwise probability value ppq in P is equal to gpq(Z, Wpq) in G, the probability that the test biological specimen has phenotypic characterization Tp and does not have phenotypic characterization Tq, wherein Z is a set of {z1, . . . , zn} cellular constituent abundance values measured from the test biological specimen for said plurality of cellular constituents; (D) optionally, instructions for converting P to a set M of k probabilities, wherein M={p1, p2, . . . , pk}, wherein each probability pj in M is a probability for a phenotypic characterization in the set of {T1, . . . , Tk} phenotypic characterizations that the biological specimen has the phenotypic characterization such that - View Dependent Claims (22, 23, 24)
-
-
25. A computer-readable medium storing a computer program executable by a computer to determine, for each respective phenotypic characterization in a set of {T1, . . . , Tk} phenotypic characterizations, a probability that a test biological specimen has the respective phenotypic characterization, the computer program comprising:
-
(A) instructions for learning a pairwise probability function gpq(X, Wpq) using a training population, for a pair of phenotypic characterizations (Tp, Tq) in the set of {T1, . . . , Tk} phenotypic characterizations, wherein; (i) there are at least five training samples in the training population for each phenotypic characterization in the set of {T1, . . . , Tk} phenotypic characterizations; (ii) Y is the set of all training samples in the training population that exhibits either phenotypic characterization Tp or phenotypic characterization Tq, and each Yi in Y is the set of {yi1, . . . , yin} cellular constituent abundance values for a plurality of cellular constituents measured from a sample i, from the training population, which exhibits either phenotypic characterization Tp or phenotypic characterization Tq; (iii) Wpq is a set of parameters derived from Y in the learning step (A) for a pair of phenotypic characterizations (Tp, Tq) by substituting each Yi into gpq(X, Wpq), as X, by the instructions for learning (A); (iv) k is 3 or greater; (v) n is at least 1; and (vi) p is not equal to q, (B) instructions for repeating the instructions for learning (A) for a different pair of phenotypic characterizations (Tp, Tq), using the training population, for all unique pairs of phenotypic characterizations in the set of {T1 . . . , Tk} phenotypic characterizations, thereby deriving a plurality of pairwise probability functions G={g1,2(X, W1,2), . . . , gk-1, k(X, Wk-1, k)}; (C) instructions for computing a plurality of pairwise probability values P={p1,2, . . . , pk-1, k}, wherein each pairwise probability value ppq in P is equal to gpq(Z, Wpq) in G, the probability that the test biological specimen has phenotypic characterization Tp and does not have phenotypic characterization Tq, wherein Z is a set of {z1, . . . , zn} cellular constituent abundance values measured from the test biological specimen for said plurality of cellular constituents; (D) optionally, instructions for converting P to a set M of k probabilities, wherein M={p1, p2, . . . , pk}, wherein each probability pj in M is a probability for a phenotypic characterization in the set of {T1, . . . , Tk} phenotypic characterizations that the biological specimen has the phenotypic characterization such that
-
Specification