Crisis Action Planning 21 January 2006 Forming the CTF UNCLASSIFIED CTF Training.
2015 Projects Sage Bionetworks. Children’s Tumor Foundation (CTF) 1. TCGA-scopre analysis of...
-
Upload
roberta-simon -
Category
Documents
-
view
213 -
download
0
Transcript of 2015 Projects Sage Bionetworks. Children’s Tumor Foundation (CTF) 1. TCGA-scopre analysis of...
2015 ProjectsSage Bionetworks
Children’s Tumor Foundation (CTF)
1. TCGA-scopre analysis of dermal NF
2. Synodos II: NF1 low grade glioma
3. CTF data portal
Goal: Assist CTF in research focused questions for NF disease, and to broadly build a community resource for CTF supported investigators
Aims
Computational Biology Team
AJCC Biomarker Project in Colon Cancer
Biomarkers: KRAS, BRAF, microsatellite instability (MSI)
Goal: Assess biomarkers for inclusion in tumor staging in CC for AJCC’s 8th edition.
Computational Biology Team
TCGA Pancan GI
Esophageal cancer
Gastric cancer
Colon cancer
Rectal cancer
Goal: Leverage existing CRCSC community for pan GI analysis
Computational Biology Team
AACR Project GENIE
3
5 4
Members
Goal: Harmonize precision medicine studies across major cancer studies
Genomics Evidence Neoplasia Information Exchange
Computational Biology Team
Accelerated Biological Aging and Colon Polyp to Cancer Progression – (with
Grady Lab)
Goal: Assess model of “accelerated biological ageing” (as previously modeled in Hannum et al) to predict malignant transformation of colon polyps into advanced adenomas.
Computational Biology Team
SPECSII Prostate Cancer
Aim 1: To validate prognostic signatures in localized, lethal PC
Aim 2: To validate pathway signatures associated with localized, lethal PC
Aim 3: To adapt prognostic and pathway signatures for clinical deployment (also untreated vs post treated)
Goal: Validation of prognostic and pathway signatures in lethal prostate cancer
Computational Biology Team
AACR Minisymposium project
Aim:Provide mechanism for AACR to assess pool of abstracts for selecting oral presentations at annual meeting.
Goals:• Cluster top rated abstracts major abstract categories - (Clinical research,
Experimental/Molecular Therapeutics, Molecular/Cellular Biology, Tumor Biology…) using terms that AACR have picked:
• Clustering of these abstracts using the interactome text mining pipeline• Build interactive visualization
Computational Biology Team
Single-Cell RO1 (with the Radich lab)
Our job: use the genomic data to infer the hierarchy of subpopulations
Research question: how to identify the degree of heterogeneity within an AML sample?
A bulk sample Sequence data
?
Sequencing
Computational Biology Team
Avon Metastatic Breast Cancer Project
International Cancer Research Partnership (ICRP) has a database of grantsProblem: everything is manual… everything is done with Excel…
Our job: develop a software tool that allows • Automatic classification, tagging, highlighting, etc.• Reusable human effort such as stored queries• Analytics using ML techniques
Q: anything related to metastatic breast cancer?A: search meta*… search metas*… search metasta*…
Q: what diseases is this grant about?A: let me read it first…
Q: what pathways is this grant about?A: … let me read it again…
Computational Biology Team
Synodos for NF-2(~12 groups)
Goal: To find genetic root causes for NF-2 and consistent/new drug treatments in a multi-institutional collar
Aims:1.Test drugs against human and mouse merlin-deficient schwannoma and meningioma cells. 2.Collect and integrate molecular systems data to prioritize existing targets, identify new target3.animal models to study efficacy signal and PK/PD for drugs that meet in vitro go/no go criteria4.Identify at least 1 drug to move forward for a clinical trial
2 prelim
manusc
ripts
Dec - 15May - 15
Sage Analysis
Aug’14
start
Sep’14 Jan’15
in-vitro
drug sc
reens
kinome
proteomics
April’15
ngs - st
ructu
ral v
ariatio
n
Animal te
sting
RNASeq
Feb’15 Sep - 15
Syndos d
ata explore
r
release
d
Progress:
Computational Biology Team
AML FederationGoal: Individualized Systems Medicine approaches to coordinated pediatric AML treatment options
Aims:• show high-throughput drug screening assays across 4 medical centers can be reproducible • predict treatment outcomes based on integrated data for each patient including treatment,
functional/biological, and genomic profiles• real-time approach to stratify patients based on tumor characteristics optimized across sites that
will then be used to provide treatment option(s) driven by international, ‘real time’ collaborative experimental and analytical decision making.
manuscrip
t
Sage Analysis
April’14
start
April’15
patient d
ata sharin
g
protoco
l in-vtiro
scre
en
data generatio
n
July’15
patient s
amples
Sep’15 Oct-Dec’15
Progress:
June’15
Computational Biology Team
AMP-AD: The Accelerating Medicines Partnership
Goal: To test whether collaboration can advance research efforts for systems biology-based identification and validation of novel drug targets in Alzheimer’s Disease.
Aims:- Centralize resources across 4 academic and 4 industry
partners- Benchmark methods across experts- Establish community consensus in results within and beyond
consortium.
Systems Biology Team
CMC: The CommonMind Consortium
Goal: To generate and analyze large scale data from human subjects with neuropsychiatric disorders through a public-private partnership, and to make this data and the analytical results broadly available to the public as a free resource that will seed community efforts for target identification in this field.
Aims:- Facilitate immediate data and resource sharing among the
CMC members and timely public release of data and analysis results
- Coordinate data generation efforts - Participate in, and drive aspects of, consortium wide
collaborative analysis
Systems Biology Team
Single Suture Craniosynostosis: Candidate Gene and Pathway DiscoveryGoal: To characterize molecular pathways leading to non-syndromic craniosynostosis using molecular data as a means to improve biological understanding of disease.
Aims:- Use transcriptome profiles to identify groups of patients with
similar molecular pathways; identify target pathway components to test.
- Confirm predictions in second dataset.- Validate targets in cell and animal based experimental
systems.
Systems Biology Team
PEC: The PsychENCODE ConsortiumGoal: To facilitate collaborative efforts to discover and characterize non-coding functional genomic elements in the human brain, and elucidate their role in the molecular pathophysiology of psychiatric disorders through unbiased genome-wide approaches.
Aims:- Facilitate immediate data and resource sharing among the PEC
members and scientific community at large- Discuss and share research strategies, approaches, technologies
and methods- Decide optimal research approaches and protocol designs- Develop uniform procedures and practices both for data
generation and analysis
Systems Biology Team
Network Inference Benchmark
Goal: To benchmark network inference algorithms and to establish a consensus toolbox for association network inference and analysis that can be used across diseases and research questions.
Aims:- Benchmark existing algorithms for network inference- Establish consensus toolbox for network inference and analysis- Application of established toolbox to understand the disease
biology of Schizophrenia and Alzheimer’s disease- Development of visual network dashboard to involve extended
research community in building networks and interpreting results.
Systems Biology Team
PCBC: Progenitor Cell Biology Consortium
Goal: Create a data portal for all molecular assays and metadata, advise and perform analysis within and across assays to create a community resource for characterization and selection of progenitor cell models for research.
Aims:-Standardize data access, annotations across assays -Establish community consensus in results within the consortium, available in Synapse -Data visualization available for interaction -Develop resources for community development from scratch
Data Science Team
LINCS (OHSU): Library of Integrated Network-based Cellular Signatures
Goal: Develop a data resource of molecular signatures from microenvironment perturbation assays, including access to and provenance of fluorescence microscopy images to establish community-consensus in research outcomes.
Aims:-Standardize data access, annotations -Data visualization available for interaction
Data Science Team
Pancan : The Cancer Genome Atlas and the International Cancer Genome Consortium
Goal: Support two communities of several hundred researchers analyzing 30+ cancers simultaneously.
Aims:-Assist in coordination to data and research as 100+ sub-projects are developed into publications.-Drive integration between PCBC and TCGA analysis for establishing stemness signals in Cancers-Drive integration between NF efforts and TCGA
Data Science Team
U54 Center of Excellence
Goal: Create a resource (with attribution) of research assets being created by researchers across the country.
Aims:● Have all research assets generated by the five cores
in the U54 represented in Synapse● Create ways for researchers to find assets and
researchers with similar research agendas● Allow junior researchers document atttribution of
reuse of their resources.
Data Science Team
Infectious Disease Resilience Challenge
Goal: Use challenge mechanism to identify natural resilience to sepsis that may be used to develop new strategies for preventative care or treatment
Aims:‑ Gather datasets to be used in challenge focused on
sepsis and create a meta challenge across those‑ Use human infectious disease challenge experiments to
crowd source the identification of signatures of resilience
Data Science Team
Prepare mHealth sensor data for feature extraction
Relevance: We need to prepare all mPower app sensor datafor extracting relevant features.
For example, we need to see how and whether removing/editing bad sensor files will improve signal.
Significance:Collecting vast amounts of mHealth data will fail to impact the way in which medical research is done unless the data are deemed of high enough quality to furnish useful findings.
Progress:We have submitted an NIH proposal. In the meantime, we are testing automated ways of assessing recording quality, and are inviting experts to process the data.
Crowdsourcing audio annotation:
ImagingTeam
Extract features from mHealth sensor data
Relevance: We need to extract relevant featuresto characterize mPower’s sensor data.
Significance:Our mHealth app-related workis intended to vastly increase the ease,scalability, and engagement of mobilehealth research applications. It is of paramount importance that the features we extract have the bestchance of drawing signal out of the data.
Progress:We are evaluating a collaborator’sfeatures and are inviting other expertsto process the data.
Imaging Team
Make questionnaires intuitive, engaging, and text-free
Relevance: We are trying to move from boring, intrusive, arduous questionnaires to passive data collection.
One example path is to translate mPower's questionnaire to visual counterparts without text.
Significance: If participants are not sufficiently engaged, all our efforts to transform participant-centered
research will be moot. The most obvious way to engage is to widen accessibility to speakers
of any language, and to visually communicate information.
Imaging Team
Visualize large image data on Synapse
Relevance: Challenge participants need to view and interact with large, server-side image data via a browser.
Significance: Image data need to be seen, especially in advance of and at intermediate steps of processing.
Our challenges will be at a disadvantage if they cannot allow participants to interact with
the image data visually.
Mike’s team: Connect vis platform with Synapse data.Thea & Arno: Coordinate evaluation.
Imaging Team
Visualize brain image data on Synapse
Relevance: Different imaging communities could be drawn together via a Synapse browser-based vis platform.
Significance: There is a great need across neuroimaging communities for a uniform way to visually navigate
imaging data, especially for quality control assessment and individual scrutiny of data.
Progress: We built a single-brain viewer prototype, and are planning a QC system for thousands of brains.
Mike’s team: Connect vis platform with Synapse data.Thea & Arno: Coordinate evaluation.
Mobile Health Study AppsAfter launching ResearchKit in March, we have opportunities to
improve data analysis, support new studies and ultimately shift how these study apps can be used for medical research
Analysis efforts have focused on mPower (Parkinson disease)
Potential partners include: patient advocacy groups, academic researchers, pharma/biotech, hospital systems, federations
We continue to explore the ability of defining states of health and disease driven by sensor phenotypes to augment standard definitions used for clinical management
mHealth
mPower mHealth research app focused on motor, gait and speech
symptoms of Parkinson disease
Analysis efforts to build a usable “scorecard” of features related to disease as a personal tracking tool of impact of drugs and other modulators of symptoms as a pilot for future work in this area
Improve the app in domains related to engagement, usability and scientific questions related to this goal as well as develop new tools that can redefine the way we diagnose Parkinson disease
mHealth
www.dreamchallenges.org
Challenge Website: https://www.synapse.org/ - !Synapse:syn2813558/wiki/March 16- July 27 Challenge Focus: Predict predict survival for prostate cancer patients based on patients’ clinical variablesOverview: This Challenge attempts to improve the prediction of survival and toxicity of docetaxel treatment in patients with metastatic castrate resistant prostate cancer (mCRPC). The primary benefit of this Challenge will be to establish new quantitative benchmarks for prognostic modeling in mCRPC, with a potential impact for clinical decision making and ultimately understanding the mechanism of disease progression.
• Challenge partner: Project Datasphere (data)• Challenge dataset: Four comparator arm data sets of Phase III prostate cancer clinical trial hosted on
the Project Data Sphere® platform• Incentives:
• Key authorship for Challenge overview paper• Invite and travel paid to 2015 DREAM Conference• NCI Model Hardening Grant of $75K• $30K in educational grants from Project Datasphere/Astrazeneca• AJCC evaluation of top models for inclusion as a Precision Medicine prognostic calculator
• Journal Partner: Nature Biotechnology
• Challenge Website: https://www.synapse.org/ - !Synapse:syn2813558/wiki/• June 23- September, 2015• Challenge Focus: Predict the progression and survival of ALS patients• Challenge Overview: Prize4Life has worked to receive, harmonize and merge more ALS clinical data
so that now there are more than 9,000 ALS patients represented in the PRO-ACT database including a large amount of previously unpublished data. The DREAM ALS Stratification Prize4Life will utilize this growth in data to address the question of stratifying ALS patients into meaningful subgroups, to enable better understanding of patient profiles and application of personalized ALS treatments. This effort will also work to significantly reduce the costs of future ALS clinical trials, aid clinical care and identify new disease predictors that can lead to novel biomarker development.
• Challenge Partners: ALS Prize4Life (data) and IBM (cloud and Docker pipeline)• Challenge Sponsors: Biogen-Idec and Lilly• Challenge dataset: PRO-ACT database of more than 8,000 ALS clinical cases, Italian and Irish registry
datasets• Incentives:
• Key authorship for Challenge overview paper• Invite and travel paid to 2015 DREAM Conference• $7,000 cash prize per sub-Challenge
• Journal Partner: Nature Biotechnology
• Challenge Website: https://www.synapse.org/ - !Synapse:syn4231880/wiki/• Opening August 2015• Challenge Focus: Predict the effect of drug combinations in cancer cell lines• Challenge Partners: Astrazeneca (data) and Sanger (data)• Challenge Sponsors: Astrazeneca• Challenge Overview: This Challenge is designed to explore – and hopefully reveal – fundamental traits that underlie
effective combination treatments and synergistic drug behavior using baseline genomic data, i.e. data collected pretreatment. At a practical level, the Challenge aims to uncover important biomarkers that are predictive of synergistic behavior, yielding a direct path for clinical translation. From the experimental perspective, the Challenge presents an opportunity to learn a set of general patterns or rules that could be used to predict synergistic behavior in new compounds or disease contexts.
• Challenge datasets: • Astrazeneca: ~30k experimentally tested drug combinations measuring cell viability over 140 drugs and 120
cancer cell lines (primarily colon, lung, and breast), and monotherapy drug response data for each drug and cell line.
• Sanger: genomic data including gene expression, somatic mutation, copy-number alterations, and methylation data
• Incentives:• Key authorship for Challenge overview paper• Invite and travel paid to 2015 DREAM Conference• Cash prize: TBD
• Journal Partner: Nature Biotechnology
•
The SMC Challenges SMC-DNA: Predict cancer-associated mutations from
WGS data SMC-Meta-pipeline: Build models to establish the
best meta-pipeline SMC-Het: Identify the best subclonal reconstruction
algorithms SMC-RNA:Predict cancer-associated mutations and
rearrangements in RNA-seq data
Somatic Mutation Calling Challenge
Open
SMC Meta-pipeline ChallengeOpen
Tumor Heterogeneity Challenge (cloud only)
Opens August 2015
RNA-Seq Fusion Calling Challenge
Opens Fall 2015
• Challenge website: https://www.synapse.org/#!Synapse:syn3157598/wiki/• Sage’s First Ideation Challenge!• June 15, 2015-January 15, 2016: (NOW OPEN!)• Challenge Focus: to use innovative approaches to identify novel pathways—including new
genes or combinations of genes, genetic variants, or sets of genomic features—involved in breast cancer susceptibility in order to generate new biological hypotheses that can be experimentally validated
• Challenge Partners: National Cancer Institute• Challenge Sponsors (funding): National Cancer Institute• Challenge datasets: Seven breast cancer GWAS data sets (phenotype, genotype and
imputation data) prepared for the Challenge and available in dbGaP• Incentives:
• Key authorship for Challenge overview paper• Invite and travel paid to 2015 DREAM Conference• Cash prize: TBD
• Journal Partner: TBD
• Challenge website: https://www.synapse.org/#!Synapse:syn4224222/wiki/• Opening: 2016• Challenge Focus: reduce the total number of recalls without increasing the number of false negative
exams• Challenge Overview: In the DM Challenge, we will crowdsource data consisting of thousands of digitized
mammograms from patients, and corresponding outcome measures in the form of post-mammogram survival duration. The goal of the DM Challenge is to apply an open science approach to develop algorithms for risk stratification of screening digital mammograms that can be used by the scientific, industrial and regulatory communities to improve breast cancer detection.
• Challenge Partners: BCSC (data) and MSSM (data)• Challenge Sponsors (funding): Arnold Foundation• Challenge datasets: correlated screening digital mammography image and clinical data• Incentives:
• Cash prize: Total purse of $1.2M• Key authorship for Challenge overview paper• Invite and travel paid to 2016 DREAM Conference
• Journal Partner: TBD
Nascent DREAM Challenges • SMC-RNA • Second Project Datasphere Challenge • mPower Ideation Challenge • Colorectal prognosis prediction Challenge • Personal Genome Project Challenge (Brian, Thea)• METS MICCAI MRI Image Challenge: conversion to a DREAM Challenge • Influenza Challenges • UK Biobank Challenge
Opening August 2015• Sage Team: Yin, Justin, Kristin, Chris, Amy, Christine, Thea• Challenge Focus: Identify the best subclonal reconstruction algorithms and to identify the conditions
that affect their performance• Challenge Partners: Ontario Institute of Cancer Research (data), Google (cloud)• Challenge dataset: series of simulated tumor genomes • Incentives:
• Key authorship for Challenge overview paper• Invite and travel paid to 2015 DREAM Conference• Winning method will be run on a subset of the ICGC pan-cancer dataset of 2500 whole-genome
sequences (subset size will depend on computational characteristics of the winning method).• Journal Partner: Nature Biotechnology• 2015 Objectives: high priority carry over objectives
• Continual Challenge to Dynamically Update Benchmarks• 2016 Objectives
• Improved ability to share running code and models. A - 2 Challenges demonstrate ability of organizers and participants to run other's models / algorithms S - Adoption of approach outside of DREAM Challenges
Platform Development Roadmap
July-Dec, 2015
Communication in Synapse
Goal: Facilitate and capture dialog within a community to increase knowledge exchange.– Refinements to Synapse Messaging– Replacement of external Get Satisfaction forum on
Synapse– Forum / Discussion area within a project– Voting on comments
Platform Team
Synapse Visualizations
Goal: Provide common visualization needs and allow non-engineers to provide for project custom needs– Thumbnails (MS Office, pdf, etc) – Scientific Visualizations : Genome Browser,
Cytoscape– Extensibility for custom needs– Longer term: medical images (2016 projects)
Platform Team
Synapse Reproducible Pipelines
Goal: Improve ability of people to re-run code / analysis supplied by others. Supports scoring on methods instead of prediction for DREAM challenges, increasing reproducibility. Supports automated data processing pipelines for other communities.– Integration with container technologies, e.g. Dockr– Provenance extensions to Container and Tables– Usability of Provenance in web
Platform Team
Bridge – Researcher Dashboard / Study Management Tool
Goal: Improve ability of researchers to directly monitor and dynamically modify their study. – Web based front-end– Standardization of data schemas across Bridge
Server, Synapse, and mobile clients
Platform Team
Key governance activities
• Qualified Users for Bridge “share broadly” data - to fulfill the promise made to study participants
• BAA white paper for website - to explain why we have not signed Business Associate Agreements under HIPAA for our work
• e-consent Methods Paper for publication - to document our work, and help others follow in our footsteps
Governance Team
supporting new uses of consent toolkit
• Multiple new studies in discussion, first “in the wild” reuse by UCSF’s Pride app - we want to catalyze many uses of visual consent and document its benefits
• Continue to create and curate more assets - icons, workflows, case studies, instructions, documentation - to make visual consent work on new classes of data and new kinds of studies.
Governance Team
design concepts!
file:///var/folders/9p/
bbpc5fkd6ybdzwss43hcfzsc0000gn/
T/noun_dna_77921.p
ng
You will spit into a tubeto provide your sample.
“Spit Kit”
next
learn more
file:///var/folders/9p/
bbpc5fkd6ybdzwss43hcfzsc0000gn/
T/noun_dna_77921.p
ng
You will send in thesample via a pre-paid
mailer.
Mailing
next
learn more
file:///var/folders/9p/
bbpc5fkd6ybdzwss43hcfzsc0000gn/
T/noun_dna_77921.p
ng
Medical Records
you’ll be asked to authorize your medical records into
the study.
Governance Team