Questions w/ Dr. Demet Sag
How do you describe yourself in terms of data science?
I see data science as an automated tool using expert knowledge as a driver to find the best path for the application of unmet health needs or forecast what is the best investment opportunity. I am an initiator, cultivator, and analyzer to get it done. Chef in the kitchen, since there is ton of data waiting to be analyzed before it can be served.
I look at the picture to find matches and relationships to the human molecular physiology to create responses to identify markers or targets. I am fortunate to have a strong industrial microbiology knowledge that leads me to observe immune responses, epidemiology, interactions, complexities of microbial physiology and integration of all for good and bad. Next, I marry my microbial organism knowledge with genetic engineering for optimizing production and reducing cost at the same time. Finally, I gear towards single cells of multi system organisms, comparative molecular development genetics, and functional genomics. I work with stellar scientists internationally: NATO, EMBO, ELBA, Japan, Canada, Germany, Switzerland, US, India, China. Translation of data for actionability is the goal.
I am a curious, adventurous, independent learner with a very diverse portfolio. I have addressed impurities and carried gained knowledge/data to further applications as lessons learned.
What I have done and am doing are interconnected with a focus on translating raw info/data into the full circle of life. Starting from a single cell I analyze genetics/genomics. My tools involve various aspects of nanotechnology and incorporate into the microbiome. At the end of the day our body has 90% microbial genes and we are using them to edit the genome and understand their defense systems. We have an established microbial flora harmonizing or competing with our immune system for a full functionality of the unit. My perspective in disease application is simplifying big problems and solving complex issues. I study how:
- cancer can be the result of a stem cell gone bad
- the cardiovascular system is a connection of vital highways
- blood is fluid of life and is the first tissue to collect snapshot of health status
- the immune system is a systemic entity instructing others systems how to react or act
Is there a simple definition of data science?
Data science is a very broad definition processing information/data using computers and creating analytical application points.
It is a method to analyze big collection of information using applied math, computer science, and expert area knowledge with a diverse group of people for artificial intelligence and optimization of solutions under the given algorithms.
How subjective is data interpretation?
Essentially data interoperation is subjective at first. Qualitative data is quantitatively mined to uncover the essential factors.
Data analysis and interpretation, regardless of method and qualitative/quantitative status, may include the following characteristics:
- Data identification and explanation
- Comparing and contrasting of data
- Identification of data outliers
- Future predictions
Quality of the data depends on who is analyzing it based on their ability to evaluate shades of gray. Obviously no machine works with 100% efficiency. Subjective data has a place to determine the next step as it is part of the process like getting a diagnosis by the physician. We provide subjective data including family info, past diseases, types of discomfort from the patient, and caregiver. All this helps to understand the problem needing to be solved. A clinician may order specific tests to identify the main method to offer treatment, which is the main objective.
How can objectivity be optimized?
Objective data analysis is a process of inspecting, cleansing, transforming and modeling data with the goal of discovering useful information, informing conclusions and supporting decision-making.
Data interpretation depends on who is asking the questions and how they are asking. Following the 60 second method as the art of asking good questions is the key. Who is analyzing the input and who is providing the output for the end user are also key. When asking someone to read a language they don’t know it is difficult to ask them to interpret context. Quality, verification, validation, and risk management are necessary to fine-tune.
Working on diverse teams with intellectual minds plays a positive role. This depends on the hierarchy in the organizational structure. Top down method is an old fashion technique; objectivity can be reached using collective thinking.
Will there ever be a “perfect” algorithm?
Everything in the universe design to reach equilibrium and homeostasis. One size doesn’t fit all since each living unit has a mine to be discovered.
The crude definition of the word algorithm relates to the data analysis procedure for a specific question in mind:
“Algorithm is a step-by-step procedure, which defines a set of instructions to be executed in a certain order to get the desired output.
- Search − Algorithm to search an item in a data structure.
- Sort − Algorithm to sort items in a certain order.
- Insert − Algorithm to insert item in a data structure.”
In life the definition of “perfect” is philosophical for some but there are basic essential rules to follow. If we accepted the earth is flat as a reference point many data would be of disservice to our survival. Even when I teach I use metaphors from daily life regarding mythology, some from today and some from the past to paint a picture. The key is estimating the best probable use of the details in the big picture.
What domain have you chosen to master?
My scientific domain aligns with the data domain for clinical genomics application. Specifically, “in data management and database analysis, a data domain is the collection of values that a data element may contain. The rule for determining the domain boundary may be as simple as a data type with an enumerated list of values.”
Example Schaeffler Data Domains contain data for full product life cycle:
- Measurement Data,
- E-Business Data,
- Linear Data,
- Manufacturing Planning Data,
- Logistic Data,
- Material Master Data,
- R&D Engineering Data,
- Project Data,
- Sales Data,
- HR Data,
- Purchasing Data,
- Finance Accounting Data
There is no picking and choosing how data works; it depends on the application. There are 10+ suggested domains we previously called “modeling.” Metaphorically, at the end of the day if you live by the sea versus the mountain or the desert, each geographical area requires different approaches. Getting from point A to B is impacted by determining which kind of transport system you are using.
Diversity and power of data increases with comparative knowledge. I use all domains to bring the idea to the market and each stage requires a different approach to reach to the next step. However, they are all regulated and documented with a logic like a puzzle: GLP, GCP, GMP, Quality, Risk, Budget, Human life protection, Human Factor Engineering/Design to avoid risks, processing SOPs, reporting AEs, following up and feeding the top to improve design etc.
Thus, Strategy, Process, Data Model, People, Roles, Responsibilities, Applications, Data Model and Data Quality align like a puzzle for an application.
The key is the strategy development and managing the project timelines.
I value alliance management for a common goal to succeed as a team. I have an experience using multiple models and diseases for clinical functional transgenomics that enables to design a well planned strategy under regulations, given budget and business model to meet the needs. Furthermore, I am an enabler to mentor people power with 20+ educator teaching experience in genomics, bioinformatics, anatomy and physiology, translational medicine, and regulation to engineers, health care providers, business majors or audiences.
My sweet tooth is cutting edge innovative technologies that requires matrix environments that creates an opportunity to learn something new directly or indirectly.
In science, attraction towards single cell development and differentiation brings precision in the clinical genetics/genomics for function.
At the end of the day the application of the data science involves interactions between genomic materials and cell-cell, cell-tissue, host-pathogen, host-tumor, or host-treatment using small or large molecule. These molecules invoke immune responses which result anytime we introduce anything foreign to our body. Safety is always first, no harm.
I focus on precision that is valuable for an effective outcome by assessing the best target(s) using small/large molecules and/or develop medical devices. Then I evaluate how we can measure success in the clinical setting when creating panels and developing medical devices.
Data is a tool and if it is not at the hands of good user the information can be lost in translation. I translate biological events to answer:
Where – (Specific tissue, cell or interaction)
When – (Child, adult, time of exposure)
How – (Synergistic activators/opposing signals, contributors, etc)
Why – (Triggers)
How do you grow your knowledge in new domains?
I follow unmet health needs and grow myself to address those needs best. I do this by: reading articles, meeting key opinion leaders, and attending events (virtually due to Covid19). Products have a life cycle and so do humans as we evolve so it is a natural process to increase fitness. My goal is to keep discovering new things along my scientific journey. Some people enjoy doing the same thing within a given niche but I am a translational clinical research scientist who likes to oversee idea until it becomes a reality. This requires many aspects like innovation, strong scientific experience, and knowledge of disease mapping. When we put these together data can be utilized to navigate our questions as a boat through both stormy and calm weather.
How do you predict data science will evolve?
There will be requirement to for data scientists to have both business savvy and knowledge of applications in life sciences. These specialties include anatomy and physiology, cellular development, stem cells, and overall system biology to make connections. At the end of the day data science is supporting how scientific evidence can be utilized with the minimum risk. We prioritize safety to select economically streamlined effective treatments.
My perspective focuses on how we can use the data, bioinformatics, and living algorithms to make treatments functional. Functional Genomics demonstrate how I can engineer pieces of information to work in human molecular physiology for the best outcome. This involves correcting the disease conditions and monitoring the health for disease prevention.