U.S. patent application number 17/512771 was filed with the patent office on 2022-02-17 for pathway visualization for clinical decision support.
The applicant listed for this patent is KONINKLIJKE PHILIPS N.V.. Invention is credited to NILANJANA BANERJEE, NEVENKA DIMITROVA, ANGEL Barnhart JANEVSKI, SITHARTHAN KAMALAKARAN, VINAY Antonius VARADAN.
Application Number | 20220050832 17/512771 |
Document ID | / |
Family ID | 1000005940603 |
Filed Date | 2022-02-17 |
United States Patent
Application |
20220050832 |
Kind Code |
A1 |
DIMITROVA; NEVENKA ; et
al. |
February 17, 2022 |
PATHWAY VISUALIZATION FOR CLINICAL DECISION SUPPORT
Abstract
When generating visual representations of gene activity pathways
for clinical decision support, a validated pathway database that
stores a plurality of validated pathways is accessed, wherein each
pathway describes at least one interaction between a plurality of
genes. A processor (18) is configured to execute
computer-executable instructions stored in a memory (16), the
instructions comprising visually representing gene activity level
(28) for at least one gene across a plurality of populations,
retrieving a pathway (32) from the validated pathway database,
wherein the pathway includes the at least one gene, and visually
representing gene activity levels for all genes in the pathway. The
instructions further comprise a formal visual grammar
representation visually representing information flow (36) through
interactions between genes in pathway, linking the visual
representation of the information flow to actionable information
(42) for a clinician, and outputting the visual representation of
the information flow and the linked actionable information to the
to the clinician.
Inventors: |
DIMITROVA; NEVENKA; (PELHAM
MANOR, NY) ; JANEVSKI; ANGEL Barnhart; (NEW YORK,
NY) ; BANERJEE; NILANJANA; (ARMONK, NY) ;
VARADAN; VINAY Antonius; (NEW YORK, NY) ;
KAMALAKARAN; SITHARTHAN; (PELHAM, NY) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
KONINKLIJKE PHILIPS N.V. |
Eindhoven |
|
NL |
|
|
Family ID: |
1000005940603 |
Appl. No.: |
17/512771 |
Filed: |
October 28, 2021 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
14390187 |
Oct 2, 2014 |
11170013 |
|
|
PCT/IB2013/052460 |
Mar 27, 2013 |
|
|
|
17512771 |
|
|
|
|
61618892 |
Apr 2, 2012 |
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G16B 5/20 20190201; G16B
45/00 20190201; G16B 5/00 20190201; G06F 16/248 20190101 |
International
Class: |
G06F 16/248 20060101
G06F016/248; G16B 45/00 20060101 G16B045/00; G16B 5/20 20060101
G16B005/20 |
Claims
1. A system that facilitates visualizing gene activity pathways,
using a visual grammar that defines visual elements associated with
gene expression, gene activity level, and information flow, for
clinical decision support, including: a validated pathway database
that stores a plurality of validated pathways, each of which
describes at least one interaction between a plurality of genes
using the visual grammar; and a processor configured to execute
computer-executable instructions stored in a memory, the
instructions comprising: visually representing gene activity level
for at least one gene across a plurality of populations; retrieving
a pathway from the validated pathway database, wherein the pathway
includes the at least one gene; visually representing gene activity
levels for all genes in the pathway; visually representing
information flow through interactions between genes in the pathway;
linking the visual representation of the information flow (36) to
actionable information for a clinician; outputting the visual
representation of the information flow and the linked actionable
information to the clinician.
Description
[0001] CROSS-REFERENCE TO PRIOR APPLICATIONS
[0002] This application a continuation of U.S. application Ser. No.
14/390,187 filed Oct. 2, 2014 which claims the benefit of
International Application No. PCT/IB2013/052460 filed on Mar. 27,
2013 which claims the benefit of U.S. Provisional Application Ser.
No. 61/618892 filed on Apr. 2, 2012. These applications are hereby
incorporated by reference herein.
[0003] The present application finds particular application in
clinical decision support systems. However, it will be appreciated
that the described technique may also find application in other
diagnostic systems, other medical scenarios, or other clinical
techniques.
[0004] Diagnostic and therapy decisions in oncology are largely
based on the underlying biology. Understanding the mechanisms of
the disease plays a key role in cancer research, but can also help
clinicians in making decisions and tracking the progress of the
disease.
[0005] Currently, next generation sequencing is close to a
revolution to provide information that was not previously
accessible to the clinicians for diagnosis and prognosis of a
particular disease. Both the depth and the resolution of this
information create an enormous amount of data. Molecular signatures
that have previously been studied typically do not explicitly
capture the underlying cellular mechanisms (pathways) and are
therefore limited in their ability to explain the real root cause
and the ability to create an understanding of how to treat the
patient within the realm of available therapy choices. Biological
pathways are an intuitive way of presenting this information.
Existing methods have focused on computing how to rank pathways and
their deregulation in the context of a particular clinical
question.
[0006] In current practice, patients diagnosed with cancer are
stratified based on clinicopathological data into groups that
determine prognosis (e.g. in terms of time to cancer progression or
recurrence), response to or selection of therapy, etc., but the
basis for stratification is typically presented as a table or a
list of markers and clinical data. Also, classification of patients
based on high throughput molecular data through the statistical
selection of a set of features that jointly differentiate between
clinically relevant classes of patients results in just a single
score or a list of genes levels. Moreover, these methods do not
explicitly capture how elements of genetic regulation may impact
the underlying condition that the doctor is trying to assess beyond
a fairly simple rule-based association, which is already present in
the clinical guidelines. However, for patients that are clearly not
falling within these boundaries of a clinical guideline there is
very little information that can be elicited from the massive
amounts of data generated by next generation sequencing.
[0007] The present application relates to new and improved systems
and methods that facilitate visually representing gene activity
levels, pathways and interactions in order to provide therapy
recommendations to a clinician, which overcome the above-referenced
problems and others.
[0008] In accordance with one aspect, a system that facilitates
visualizing gene activity pathways using a visual grammar that
defines visual elements associated with gene expression, gene
activity level, and information flow, for clinical decision support
includes a validated pathway database that stores a plurality of
validated pathways, each of which describes at least one
interaction between a plurality of genes, and a processor
configured to execute computer-executable instructions stored in a
memory. The instructions comprise visually representing gene
activity level for at least one gene across a plurality of
populations, retrieving a pathway from the validated pathway
database, wherein the pathway includes the at least one gene, and
visually representing gene activity levels for all genes in the
pathway. The instructions further comprise visually representing
information flow through interactions between genes in pathway,
linking the visual representation of the information flow to
actionable information for a clinician, and outputting the visual
representation of the information flow and the linked actionable
information to the clinician.
[0009] In accordance with another aspect, a non-transitory
computer-readable storage medium has stored thereon
computer-readable instructions for visualizing gene activity
pathways using a visual grammar (230) that defines visual elements
associated with gene expression using a visual grammar that defines
visual elements associated with gene expression, gene activity
level, and information flow, for clinical decision support, the
instructions comprising visually representing gene activity level
for at least one gene across a plurality of populations, retrieving
a pathway from a validated pathway database, wherein the pathway
includes the at least one gene, and visually representing gene
activity levels for all genes in the pathway. The storage medium
further comprises instructions for visually representing
information flow through interactions between genes in pathway,
linking the visual representation of the information flow to
actionable information for a clinician, and outputting the visual
representation of the information flow and the linked actionable
information to the clinician.
[0010] According to another aspect, a method of visualizing gene
activity pathways for clinical decision support using a visual
grammar (230) that defines visual elements associated with gene
expression, gene activity level, and information flow comprises
visually representing gene activity level for at least one gene
across a plurality of populations, retrieving a pathway from a
validated pathway database, wherein the pathway includes the at
least one gene, and visually representing gene activity levels for
all genes in the pathway. The method further comprises visually
representing information flow through interactions between genes in
pathway, linking the visual representation of the information flow
to actionable information for a clinician, and outputting the
visual representation of the information flow and the linked
actionable information including at least one therapy plan to the
clinician.
[0011] Still further advantages of the subject innovation will be
appreciated by those of ordinary skill in the art upon reading and
understanding the following detailed description.
[0012] The innovation may take form in various components and
arrangements of components, and in various steps and arrangements
of steps. The drawings are only for purposes of illustrating
various aspects and are not to be construed as limiting the
invention.
[0013] FIG. 1 illustrates a system that facilitates providing a
clinician with a visual representation of a gene activity pathway
for clinical decision support.
[0014] FIG. 2 illustrates a method of visually or graphically
representing to a user or clinician a single clinical value (e.g.,
gene activity or the like) to a three-dimensional (3D) space.
[0015] FIGS. 3A-3D illustrate examples of a visual grammar in which
each node in a pathway is represented by a visual element using
color, texture, shape, etc., to represent high-throughput molecular
profiling data from single patient or a subset of patients in the
context of specific biological node (e.g., gene, locus, etc.).
[0016] FIG. 4 illustrates a mapping of gene activity.
[0017] FIG. 5 illustrates a mapping represented with a temperature
graph.
[0018] FIG. 6 illustrates a visualization of multiple genes within
a pathway and association of a gene with good or poor outcome.
[0019] FIGS. 7A and 7B illustrate visual representations of an
information flow through the interactions amongst genes captured in
the pathway.
[0020] FIG. 8 illustrates visual representation that omits the node
representation and emphasizes the information flow throughout the
pathway.
[0021] FIG. 9 illustrates an example of a pathway visual
grammar.
[0022] The subject innovation overcomes the aforementioned problems
by addressing a clinical need for improved diagnostics with
visualization tools for high-throughput molecular profiling data in
general, and DNA sequencing data in particular. The described
systems and methods aid in visualizing of the statistical analysis
results in the context of pathway evaluation to stratify cancer
patients with high sensitivity and specificity, which results in
better patient outcomes, more targeted treatment, and substantial
savings in medical cost.
[0023] There is an overwhelming amount of molecular information
becoming available that can be used for diagnostic and therapy
planning purposes. The modalities span DNA sequencing,
transcriptome sequencing, methylation sequencing, etc. Providing
clear and intuitive view of this information and ability to
interact with this information is technically challenging. The
described systems and methods facilitate visualizing this
information in a specific biological pathway using high throughput
molecular profiling data derived from human samples in order to
support prognosis and therapy selection in cancer and other
diseases. Known biological knowledge about gene function and gene
interactions is captured as biological pathways, on which are
overlaid the activity levels of genes as measured from multiple
molecular modalities, such as copy number and gene expression data.
Visualization of various parameters is facilitated, including but
not limited to: individual genes within a single patient; pathway
information in a population of patients; pathway information across
different clinical studies or across patients from different
hospitals or across different regiments of pathway activity levels
in patients and these pathway activity levels can then be used to
differentiate one patient from another; pathway information
representing response to a particular therapy regimen; etc. In
another embodiment, additional clinical data associated with the
patient is provided and accompanies the gene activity and pathway
information. For example, if metabolic pathways are being
visualized, then it may be pertinent to include clinical
information indicating that the patient is diabetic.
[0024] The described systems and methods can be used in conjunction
with a pathway evaluation model such as a pathway information flow
model. In this manner the clinician is provided with a technique to
interpret whole genome molecular profiling data acquired from this
patient, to choose a much more tailored therapy for the patient. In
effect, massive and complex amounts of data are converted into
reduced visual representations that are more intuitively
interpretable by the clinician, to guide the clinician's choices
thereby significantly improving patient response while reducing the
overall costs and toxicities for the patient.
[0025] FIG. 1 illustrates a system 10 that facilitates providing a
clinician with a visual representation of a gene activity pathway
for clinical decision support. The system 10 includes a genetic
information database 12 that includes gene activity data for a
plurality of patients including a current patient, as well as gene
activity data for one or more normal patient populations and one or
more tumor patient populations. A workstation 14 (e.g., a computer
or the like), includes a memory 16 (i.e., a computer-readable
storage medium) that stores computer-executable instructions, and a
processor 18 that executes the computer-executable instructions to
perform various methods, techniques, functions, and the like as
described herein. In one embodiment, the memory is distributed
and/or provided as a service, such as a cloud-based storage service
wherein pathways and/or associated information are stored in the
cloud. The system further comprises a visualization tool 20
including an input device 22 (e.g., a keyboard, mouse, stylus,
microphone, touchscreen, touchpad, or the like) via which a user or
clinician inputs information into the workstation, and a display on
which information (e.g., visualizations or mappings, therapy
recommendations, etc.) is presented to the user. In one embodiment,
the visualization tool and/or input device are installed or
otherwise provided as an app on a smartphone, tablet, or other
personal computing device.
[0026] The memory stores a visual representation module 26 that,
when executed by the processor, generates visual representations
that are intuitively interpretable by the clinician. For instance,
the visual representation module generates a gene activity level
mapping 28 that shows a current patient's gene activity level
relative to an average activity level for the gene in a tumor
population and an average activity level for the gene in a normal
population. Once gene activity level has been visualized, the
processor retrieves a validated pathway for the gene is retrieved
from a validated pathway database 30, and the retrieved pathway is
stored in the memory 16. The visual representation module 26
generates a pathway activity level mapping 36 that shows activity
levels of all genes in the retrieved pathway, and an information
flow diagram 36 that shows relationships between the genes in the
pathway. Once the information flow diagram is generated, a linking
module 38 is executed by the processor to identify one or more
therapy plans in a therapy plan database 40 that can be used to
treat the patient based on the information contained in the
mappings 28, 34 and the information flow diagram 36. In one
embodiment, therapy plans in the therapy plan database are tagged
(e.g., using metadata) as corresponding to one or more particular
pathways and/or information flows. In another embodiment, the
therapy plan database 40 and/or the memory 16 includes a therapy
plan lookup table (LUT) that correlates therapy plans to mappings
and/or information flows. One or more identified linked therapy
plans is stored in the memory 16. The processor 18 outputs the gene
activity level mapping 28, the pathway activity level mapping 34,
the information flow 36, and the linked therapy plan(s) to the
clinician via the display 24.
[0027] As stated above, the system 10 includes the processor 18
that executes, and the memory 16 that stores, computer-executable
instructions (e.g., routines, programs, algorithms, software code,
etc.) for performing the various functions, methods, procedures,
etc., described herein. Additionally, "module," as used herein,
denotes a set of computer-executable instructions, software code,
program, routine, or other computer-executable means for performing
the described function, or the like, as will be understood by those
of skill in the art.
[0028] The memory may be a computer-readable medium on which a
control program is stored, such as a disk, hard drive, or the like.
Common forms of non-transitory computer-readable media include, for
example, floppy disks, flexible disks, hard disks, magnetic tape,
or any other magnetic storage medium, CD-ROM, DVD, or any other
optical medium, RAM, ROM, PROM, EPROM, FLASH-EPROM, variants
thereof, other memory chip or cartridge, or any other tangible
medium from which the processor can read and execute. In this
context, the systems described herein may be implemented on or as
one or more general purpose computers, special purpose computer(s),
a programmed microprocessor or microcontroller and peripheral
integrated circuit elements, an ASIC or other integrated circuit, a
digital signal processor, a hardwired electronic or logic circuit
such as a discrete element circuit, a programmable logic device
such as a PLD, PLA, FPGA, Graphical card CPU (GPU), or PAL, or the
like.
[0029] The described systems and methods facilitate clinical
decision support visualization that uses known biological pathways
and/or inferred regulatory networks to highlight pathway (network)
activity in the context of understanding the clinical condition of
a patient and/or in the context of contrasting different patient
groups. These techniques employed multiple high throughput
molecular modalities such as gene expression and copy number data
measured on the same patient sample.
[0030] The clinical decision support aspect of these techniques
involves the use of multi-modality molecular profiling data on a
single patient along with pathway database resources (for example,
National Cancer Institute Pathway Interaction Database) and a
pathway visualization engine. In this manner, an intuitive and
accurate visual representation of gene activity is generated in a
consistent manner that is supported on the technical side with a
"visual grammar" that can express the deviation from normal
activity of a gene in the context of a biological network (or a
pathway). Thus a new diagnostic device is provided that is
supported by a software and/or a hardware implementation (for
robustness and speed in a clinical setting).
[0031] FIG. 2 illustrates a method of visually or graphically
representing to a user or clinician a single clinical value (e.g.,
gene activity or the like) to a three-dimensional (3D) space. The
3D space may be a color space (e.g., RGB, CMY, or the like), a 3D
space that is presented to the user as a textured or patterned 3D
graph or representation, or any other suitable 3D space. The method
can be applied to both biological pathways and biological networks
and implements a visual grammar that can translate activity of each
gene and pathway into a visual representation. The visual
representation can use color, texture, shape, and/or temporal
aspects (e.g., flashing or motion). In one embodiment, the function
of a gene and/or the genomic locus (e.g. transcription factor that
regulates the activity of many genes or a non-coding RNA that can
keep the level of gene expression at a certain level or disrupt
activity of many genes) is visually represented such that the
frequency of the flashing or motion is proportional to the gene
activity level. Additionally, the method facilitates modularization
in that a smaller part of a network can become a single node at the
level of interacting pathways. The method also facilitates guiding
choices in clinical decision support by performing visualization of
pathways that are functionally relevant from a clinician's point of
view, for instance: per subtype of cancer, which are most likely
pathways that are varying within a particular subtype; and/or by
highlighting pathways that are potentially interesting for a
specific set of regimens (e.g., clinical decision choices) relevant
from the clinical guideline point of view. The suggested
visualizations are supported by sufficient clinical evidence. For
example, if one of the drugs in the regimen has anti-angiogenesis
action, it is clinically relevant to visualize pathways that are
related to vascular development.
[0032] Accordingly, the method comprises accessing a visual grammar
library or the like, at 78. The method further comprises visually
representing a gene activity (or some other parameter) level (e.g.,
on a graphical display or printed sheet) to a user via a
user-selected visual grammar, at 80. At 82, a relevant pathway is
obtained from a database of validated pathways. At 84, overall
pathway activity is represented visually. At 86, information flow
through the interactions amongst genes captured in the pathway is
represented. At 88, the foregoing visual representations are linked
to actionable information for a clinician. At 90, the visual
representations are presented to the clinician for analysis. The
foregoing acts are described in greater detail with regard to FIGS.
3A-9.
[0033] FIGS. 3A-3D illustrate examples of a visual grammar in which
each node in a pathway is represented by a visual element using
color, texture, shape, etc., to represent high-throughput molecular
profiling data from single patient (or a subset of patients in the
context of specific biological node (e.g., gene, locus, etc.). The
visual element may be, for example a color-coded or texture-coded
circle, rectangle (or any desired shape), a temperature map, pie
chart, bar plot, mathematical function (e.g. survival curve), or
the like. In FIG. 3A, a representation 100 is provided of the
overall mean gene expression value of a tumor population 102 and a
normal population 104. In the middle is shown the value of gene
expression value of a current patient 106. In FIG. 3B the same gene
expression values 102, 104, 106 are represented using a color-coded
rectangular node 110. Each of the visual elements in FIGS. 3A, 3B,
3C, 3D is a visual element representing a gene, its expression
status for a patient, and at least one other patient population
(e.g., normal and tumor, or two subtypes (e.g. luminal and basal
subtypes in breast cancer, note that 3D is representing only one
patient within a single patient population).
[0034] In FIG. 3C a representation 120 of the gene expression
values 102, 104, 106 is illustrated with different textures or
patterns representing the normal population, tumor population, and
the patient, respectively. FIG. 3D shows a representation 130 that
employs a temperature map 132 that visually expresses the range of
values and where the expression value of the current patient falls
within the range of values.
[0035] In one embodiment, the overall tumor and normal populations
includes all such patients within a hospital network or the like,
and the current patient may be a single patient or a subset of
patients exhibiting particular clinical variables. Additional
concentric circles (or rectangles, etc.) can be added to represent
multiple layers of patient stratification (e.g. responders and
non-responders within the tumor group) to a particular drug. In
another embodiment, a color level is associated with intensity of
expression for a positive outcome vs. poor outcome.
[0036] FIG. 4 illustrates a mapping (e.g., a visual representation)
140 of gene activity, such as is generated at 80 (FIG. 2) and as
shown in FIG. 9. The mapping is a visualization of color-based,
high-throughput molecular profiling data from single patient in the
context of specific biological pathway relative to a population.
The mapping is generated from a set of existing expression values
from one or more of: SNP data (e.g. number of mutations within a
gene), copy number polymorphisms (number of copies), gene
expression level, protein expression level, DNA methylation level
of a genetic locus, histone methylation level of a genetic locus,
histone acetylation level of a genetic locus, phosphorylation
signal (on or off) of a particular protein, etc. The first step is
to visually represent the level of activity of each gene: using
multiple molecular modalities such as gene expression data and copy
number variation data, a probability is determined that a
particular gene is over-expressed, under-expressed or is at
baseline levels compared to normal tissue. For this step, inferring
the level of a gene can be performed using predetermined thresholds
with respect to the mean of the distribution (one could also use a
simple statistical test) and the gene can be represented via one of
the several options.
[0037] Under a rule-based approach, maps a range of values from the
under-expressed to one or more of: a single color (e.g., blue) and
another range of values from the over-expressed to another color
(e.g., red); a single texture; a combination of color and texture;
and a temporal element that slightly changes the color over time
(e.g., pulsates) or the shape used to represent the carious
population values (e.g., in cases where the gene is highly
variable).
[0038] In another approach, visual color is represented in the
hue-saturation-brightness (HSB) space corresponding to an intensity
(level) of gene expression, which is generated by a continuous
function mapping. In one embodiment, if a gene has a range of
values: r=[a,b] then values are assigned in the HSB space. For each
point r.sub.i.di-elect cons.[a,b] a linear function is chosen to
map the value in h.sub.i=f(r.sub.i); however, this function can
also be a quadratic or logarithmic function depending on the
dynamic range of values. In another embodiment, when there are
functional designations of the genes such as, e.g., tumor
suppressors and oncogenes, a different slice of the HSB space is
assigned to these different functional categories. For instance,
hue is mapped to values from, e.g., 0 to 60 for oncogenes and
mapped to values from, e.g., 120 to 180 for tumor suppressors.
[0039] FIG. 4 shows an instance of a possible visualization of the
same pathway where for each gene there are multiple circles
representing a given node with respect to the mean intensity values
of the tumor population 102, mean intensity values of the normal
population 104, and the intensity values for a patient 106 (e.g.,
John Doe), respectively. The overall pathway level 142 is
represented with a different visual element (in this case a
rectangle).
[0040] FIG. 5 illustrates a mapping 160 represented with a
temperature graph that shows an overall range of values 162, a
vertical line that corresponds to the tumor population mean
intensity value 164 (e.g., for a tumor population or for a normal
population), and a dot representing the current patient's intensity
value(s) 166. The mapping is a visualization of color-based,
high-throughput molecular profiling data from single patient in the
context of specific biological pathway relative to continuum of a
population. This representation can be used for each gene in the
pathway and for the overall activity of the pathway. It should be
noted that the overall activity of the pathway may reflect the
p-value that is derived from the overall activity assessment (e.g.,
with a hypergeometric test) or derived from the information
flow.
[0041] FIG. 6 illustrates a visualization 180 of multiple genes
within a pathway and association of a gene with good or poor
outcome. The visualization shows color-based, high-throughput
molecular profiling data from single patient 182 in the context of
specific biological pathway with regard to survival of patient
groups. Each gene may individually be associated with good or poor
outcome. However, the overall synergistic effect within a pathway
could be more strongly associated with a particular outcome, and so
the overall activity of a pathway (normal vs. disrupted) is
visualized with respect to outcome.
[0042] FIGS. 4-6 thus illustrate mappings or visualizations of
overall activity of a given pathway, as is performed at 84 (FIG. 2)
once the pathway is obtained from the validated pathway database
(at 82). In one embodiment, rule-based visualization is performed
whereby a range of values is mapped from the inactive genes,
moderately active genes, and hyperactive genes, compared to normal
tissue. In one example, a single color (e.g., blue) is used for the
inactive genes, and another color is used for over-expressed genes
(e.g., red). In another example, a single texture is used for each
of the inactive genes, moderately active genes, and hyperactive
genes. In another example a combination of color and texture is
used to represent different gene activity levels. If the pathway
level is highly variable, a temporal element can be employed that
slightly changes the color over time (e.g., pulsates) or the oval
(or other shape) used to represent the patient, tumor, and normal
populations respectively.
[0043] In another embodiment, the mapping is generated using visual
color in the HSB space corresponding to respective intensities of
pathway activity levels, and is generated by a continuous function.
For instance, the input to this continuous function can be the
p-value after a hypergemetric test is applied. The input can also
correspond to the information flow level. For example, if a pathway
activity has a range of values such that: r=[a,b], then values are
assigned in the HSB space (i.e., hue-saturation-brightness). For
each point r.sub.i.di-elect cons.[a,b], a function is selected to
map the value in h.sub.i=f(r.sub.i). It will be appreciated that
the function need not be linear function, but can also be a
quadratic or logarithmic function depending on the dynamic range of
values.
[0044] Additionally or alternatively, if there are functional
designations of the pathways (e.g., signaling cascade vs.
transcription activation), there is a different slice of the HSB
space assigned to these different functional categories. For
instance, hue is mapped to values from, e.g., 0 to 60 for
transcription activation, and mapped to values from, e.g., 120 to
180 for signaling, where hue is the angle around the central
vertical axis in the HSV cylindrical space. Brightness corresponds
to the intensity of the signal.
[0045] FIGS. 7A and 7B illustrate visual representations 200, 202,
respectively, of an information flow through the interactions
amongst genes captured in the pathway, as is performed at 86 (FIG.
2). For example, a pathway is assumed that defines an interaction
between GeneA, GeneB and GeneC as shown in FIG. 7A. The interaction
in the pathway says that either GeneA or GeneB is over-expressed in
order for GeneC to be over-expressed. Using the probabilities of
over-expression, repression, and baseline expression for GeneA,
GeneB, and GeneC in the patient, a determination is made regarding
whether this particular interaction was activated. In other words,
the probability of this interaction (I.sub.1) is the probability of
GeneB or GeneA being over-expressed. A different color is assigned
for the arc between GeneA and GeneB based on the probability of
this link being active. If there is a high probability that gene C
is active and geneA is active then a strong first color is assigned
to the interaction there between (green in the example of FIG. 7B).
If the geneC is active but geneB is inactive, then we assign a
strong second color (e.g., red) to the interaction between genes B
and C.
[0046] FIG. 8 illustrates visual representation 220 that omits the
node representation and emphasizes the information flow (the
arrows) throughout the pathway 222. The information flow can now be
considered as a series of modular units in all subsequent analysis.
In one embodiment, a graph rewriting method is employed, which
consists of a set of a graph rewrite rules in the form L.fwdarw.R
where the L represents the pattern side and R is the replacement
graph. In large pathways this is useful in order to see the true
connections among key regulatory genes. As in the case of a visual
grammar, for each application, different L.fwdarw.R rules are
applied. As in the example shown in FIG. 8, the originating nodes
of all the weak connections (or disrupted connections) are deleted
thereby keeping the strongest information flow among the genes in
the pathway.
[0047] Once the information flow is visualized, a visualization
tool (FIG. 1) linked to the actionable information from the
standpoint of the clinical oncologist is employed to facilitate
clinical decision support. The clinical decision support system
(FIG. 1) links the visualization of the pathway activity to a set
of therapies that facilitate disrupting the activity of the
pathway, acting on the key readout genes directly, etc. In some
cases, the mechanism of action may not have been confirmed with
detailed biological experiments. In that case, an association
between information flow and therapy options can be made with a
predefined set of labels describing the functional aspects of
pathway activity. In this manner, the described systems and methods
facilitate providing oncology decision support such as prognosis
and therapy-response. Visual identification of and presentation of
pathways with altered activity levels can be used for developing
more individualized regimens for the cancer patients.
[0048] FIG. 9 illustrates a pathway visual grammar 230 example,
which is a context free grammar defined by a (i) starting symbol
(pathwayChart), (ii) terminal vocabulary (square, triangle,
pentagon, arrow, denim texture, red color), (iii) non-terminal
vocabulary (connectedNodes, node, nestedGeneShape, geometricObject,
shape, color, texture, connector, molecularFuncShape) and (iv) set
of productions (or rules). For instance, the pathway grammar
includes starting with a pathway chart, and identifying a number of
connected nodes in the pathway. The connected nodes are then
concatenated in a node-connector-node format. Each node is then
represented as a nested gene shape, which is further represented as
either a geometric object or a molecular function shape. The
geometric object is represented as a square shape in this example,
with one or more of color and texture superimposed thereon.
Connectors between nodes are also visually represented (e.g., as
arrows or the like). Molecular function shapes such as
transcription factors or non-coding RNA is represented by a
pentagon and a triangle, respectively, in the example of FIG.
9.
[0049] The innovation has been described with reference to several
embodiments. Modifications and alterations may occur to others upon
reading and understanding the preceding detailed description. It is
intended that the innovation be construed as including all such
modifications and alterations insofar as they come within the scope
of the appended claims or the equivalents thereof.
* * * * *