Daniel V Platt

US Patent:

6349265, Feb 19, 2002

Filed:

Mar 24, 1999

Appl. No.:

09/275535

Inventors:

Michael C. Pitman - Pine Bush NY
Daniel E. Platt - Bedford Hills NY

Assignee:

International Business Machines Corporation - Armonk NY

International Classification:

G06F 1900

US Classification:

702 27, 702179, 702181

Abstract:

The method of the present invention transforms descriptor vectors that characterize molecular complexes partitioned into groups into a space that discriminates between those groups in a well defined optimal sense. First data is generated that represents a differences between the groups of descriptor vectors. Second data is generated representing variation within the groups of descriptor vectors. A set of component vectors is then identified that maximizes an F distributed criterion function that measures differences of desciptor vectors between groups relative to varations of descriptor vectors within groups. A statistic is generated for subsets of the component vectors. For each particular subset of component vectors, a probability value for the statistic associated with the particular subset is calculated. The subset with the minimum probability value is selected. Finally, one or more of the descriptor vectors for the molecular vs complexes are mapped to a space corresponding to the selected subset of component vectors.

US Patent:

6408321, Jun 18, 2002

Filed:

Mar 24, 1999

Appl. No.:

09/275158

Inventors:

Daniel E. Platt - Bedford Hills NY

Assignee:

International Business Machines Corporation - Armonk NY

International Classification:

G06F 738

US Classification:

708520, 708400, 702 19, 702 22

Abstract:

The method of the present invention transforms descriptor vectors that characterize items partitioned into groups into a space that discriminates between those groups in a well defined optimal sense. First data is generated that represents a differences between the groups of descriptor vectors. Second data is generated representing variation within the groups of descriptor vectors. A set of component vectors is then identified that maximizes an F distributed criterion function that measures differences of descriptor vectors between groups relative to variations of descriptor vectors within groups. A statistic is generated for subsets of the component vectors. For each particular subset of component vectors, a probability value for the statistic associated with the particular subset is calculated. The subset with the minimum probability value is selected. Finally, one or more of the descriptor vectors for the items are mapped to a space corresponding to the selected subset of component vectors.

US Patent:

7356415, Apr 8, 2008

Filed:

Mar 24, 1999

Appl. No.:

09/275568

Inventors:

Michael C. Pitman - Pine Bush NY, US
Daniel E. Platt - Bedford Hills NY, US

Assignee:

International Business Machines Corporation - Armonk NY

International Classification:

G01N 33/48

US Classification:

702 19

Abstract:

A method in a data processing system for generating and storing in a database at least one descriptor vector and at least one reference frame for at least one region of a molecule includes a step of generating an entry including: i) a key derived from the at least one descriptor vector, wherein the key identifies the at least one region, and ii) a set of axes derived from property distribution information of the at least one region, the set of axes characterizing the at least one region. The method further includes steps of applying a mapping to the descriptor vector associated with the at least one region based on preselected criteria and storing the entry in a memory, wherein the key is associated with the entry such that the key indexes the entry for retrieval thereof.

US Patent:

57842949, Jul 21, 1998

Filed:

Jun 9, 1995

Appl. No.:

8/489271

Inventors:

Daniel Enoch Platt - Bedford Hills NY
Benjamin David Silverman - Millwood NY

Assignee:

International Business Machines Corporation - Armonk NY

International Classification:

G06F 1546

US Classification:

364496

Abstract:

A computer-based method and system describes molecules in a most fundamental and compact way using a set of attributes of the molecule derived from data representing the atomic structure and atomic charge of the molecule. The attributes include the shape of the molecule as defined by the moment of inertia of the molecule, the charge distribution of the molecule as defined by a novel representation of molecular quadrupole, and/or attributes that represent the relationship of the shape to the charge distribution of the molecule. A set of these physical attributes are represented by a set of descriptors. The set of descriptors may be used for molecular matching and activity prediction, as well as in 3D-QSAR analysis.

US Patent:

20190102513, Apr 4, 2019

Filed:

Oct 4, 2017

Appl. No.:

15/724377

Inventors:

- Armonk NY, US
Daniel E. PLATT - YORKTOWN HEIGHTS NY, US

International Classification:

G06F 19/22
G06F 19/18
G01N 33/48
C12Q 1/68

Abstract:

Embodiments of the present invention are directed to a computer-implemented method for generating a list of genetic variants. A non-limiting example of the computer-implemented method includes receiving genetic and biological data. The exemplary method also includes generating data patterns from the genetic and biological data with data mining. The method also includes determining redescription distances between each of a plurality of data patterns. The method also includes generating computational homology filtrations from the redescription distances using a topological data analysis and homology groups including homology group elements based upon the computational homology filtrations. The method also includes generating a single nucleotide polymorphism combination list based upon the homology group elements and redescription clusters.

US Patent:

20190102514, Apr 4, 2019

Filed:

Nov 2, 2017

Appl. No.:

15/801694

Inventors:

- Armonk NY, US
Daniel E. PLATT - YORKTOWN HEIGHTS NY, US

International Classification:

G06F 19/22
G06F 19/18
C12Q 1/6874
G01N 33/48

Abstract:

Embodiments of the present invention are directed to a computer-implemented method for generating a list of genetic variants. A non-limiting example of the computer-implemented method includes receiving genetic and biological data. The exemplary method also includes generating data patterns from the genetic and biological data with data mining. The method also includes determining redescription distances between each of a plurality of data patterns. The method also includes generating computational homology filtrations from the redescription distances using a topological data analysis and homology groups including homology group elements based upon the computational homology filtrations. The method also includes generating a single nucleotide polymorphism combination list based upon the homology group elements and redescription clusters.