BIOT 1012: Research Methods

• BIOT 1012 is a course on Research Methods and Data Management
• The lecturer is Dr. Mohamed Samir, and his email is [email protected] Office location: G218
• Lecture 1 Date: 23.01.2025

Topics covered:

• Experimental design
• Types of hypotheses
• Models in research
• Experiments versus studies
• Components of experiments/study
• Types of experiments
• Types of studies
• How data analysis, management, and statistics are affected by experimental design
• A Research cycle flows from identifying an idea through identifying biomarkers as the goal.
• Key steps in a research cycle:
  • Identifying a promising biomarker to study, which is collected from subject groups from blood samples
  • Send sample to Mass Spectrometry
  • Analyze data with machine learning
  • Identify biomarker

Hypothesis

• Hypothesis generating/free:
  • Aims to identify potential metabolite biomarkers for COVID-19 severity
  • Collect blood plasma samples from participants with COVID-19 and perform untargeted metabolomic analysis
  • Hypothesis is not required
  • Data output is large
  • Aims at patterns/relations
  • Novelty with novel molecules
  • Examples such as Omics
• Hypothesis-driven:
  • Aims to test the role of specific lipid metabolites as biomarkers for inflammatory response in COVID-19
  • Conducts blood samples from COVID-19 patients and non-infected controls for lipid molecules confirmation
  • Hypothesis is required
  • Data output is limited
  • Aims to confirm/reject hypothesis
  • Novelty is known
  • Examples like is Drug A effective than Drug B

Models in Research

• Common models:
  • In-vivo in a whole living biological system
  • In-vitro in a laboratory
  • Ex-vivo on parts/tissue from living biological systems
  • In-silico on a computer
• In-vivo:
  • Is carried out in a whole living biological system
  • Examples of models are CAM, drosophila, zebrafish, rodent, dog, pig, primates
  • Mice, rat, human, pet, monkeys, rabbits, and camels can be used
• In-vitro:
  • Is carried out in a laboratory.
  • Uses primary cells in cell culture (2D) and a test tube
• Ex-vivo:
  • Is carried out on parts/tissue from a living biological system.
  • Uses organs in a complete tissue
  • An example is human lungs transferred to a lab
• In-silico:
  • Is carried out on a computer
  • Examples include molecular docking, bioinformatics, and machine learning.

In Vivo, In Vitro, Ex Vivo

• In-Vivo:
  • Research conducted on living organisms.
  • High complexity that includes interactions between various biological systems.
  • Limited control over variables due to being a complex system.
  • Involves ethical concerns.
  • Aims to translate to human biology directly.
  • Applications include drug testing, disease modeling, genetic/behavioral studies.
  • Time scale is long
  • Animal and human clinical trials
• Ex-Vivo:
  • Research using tissues, organs, or cells taken from a living organisms and maintained outside the body.
  • Moderate complexity
  • Greater control over variables allowing for more regulated conditions.
  • Involves less ethical issues
  • May lose functional complexity
  • Application for tissue function studies, drug testing, metabolic studies and transplants
  • Moderate time scale
  • Isolated human or animal tissues
• In-Vitro:
  • Research using isolated cells or cell cultures in a laboratory setting.
  • Low to moderate simplified model with individual cells or cell cultures.
  • Highest control over variables.
  • Ethical concerns are low
  • May not fully replicate in-vivo environments
  • Application for drug screening, genetic and cellular response studies
  • Time Scale is short
  • Cell culture models
• Choice of Models Depends on:
  • Research place (resources or funds)
  • Ethical considerations
  • Research question (transmission studies need in-vivo systems)
  • Level of biological complexity
  • Time needed
  • Confounders
• In-Vivo is best for whole-organism systemic processes and complex interactions
• Ex-Vivo is best for isolated organs or tissues with more control but preserves biological complexity
• In-Vitro is best for cellular or molecular mechanisms in a simplified environment with maximum control
• Considerations when using In-vivo models:
  • Biological relevance and transmissibility - for example, the use of ferrets vs monkeys
  • Relevance to disease - for example, ducks and fish
  • Size and practical considerations
  • Genetic and Physiological Features
  • Species-specific characteristics like the immune system and metabolism

Experiments vs Studies

• Estrogen treatment in post-menopausal women was studied.
• Observational study:
  • 93,676 enrollees started in 1991 and were tracked for 8 years.
  • The input variables were whether they used estrogen treatments and factors like age, race, and diet.
  • The output variable was coronary heart disease and breast cancer
  • Positive results: increased likelihood of good health associated with estrogen treatments
• Experimental study:
  • The Women's Health Initiative(WHI) randomized a controlled trial.
  • 373,092 were eligible, 18,845 consented, and 16,608 enrolled
  • Used 16,608 women randomized to estrogen or no estrogen.
  • Women blocked together by age (3 groups; 50-59, 60-69, and 70-79), and by clinic attended and the treatments randomly assigned
  • Positive results that increased CHD and breast cancer association with estrogen treatment
• Observational Experiment is different:
  • Due to its subjects already deciding to take estrogen before the study started
  • Its cohort being more health conscious
  • A correlation being due to participants taking better care of themselves
• Observation vs Experiment:
  • Observational can establish association but not causation
  • Experiment can establish both.
• Experiment vs Study:
  • Data source
    • Designed Experiment: Researchers
    • Observational Study: Exists beforehand
  • Randomized treatments
    • Designed Experiment: Yes
    • Observational Study: No
  • Surrounding conditions
    • Designed Experiment: Controlled conditions (matched groups); and no error/bias or confounders
    • Observational Study: not controlled and has a confounder effect
  • Conclusions
    • Designed Experiment: Association & Causation
    • Observational Study: Association only
  • Within unit variation
    • Designed Experiment: Low
    • Observational Study: High
  • Solid conclusions
    • Designed Experiment: Need for few units
    • Observational Study: Needs high number of units
  • Disadvantage
    • Designed Experiment: Tight control; Not a relevant model and non-applicable
    • Observational Study: More variation
  • Advantage
    • Designed Experiment: Needs repetition
    • Observational Study: Real data applicable
  • Type of tests
    • Designed Experiment: Univariate or bivariate
    • Observational Study: Multivariate

Aspirin Experiment

• Aspirin significantly prevents heart diseases, which is an experiment
  • S1 through S10
  • Myocardial infarction, stroke and deaths were measured

Components of Experiments/Study

• Research question "Which drug has more effect on blood pressure?"
• Three sets for each test
  • Control w/o drugs
  • Drug A
  • Drug B
• The experiment can be measured at the end of Day 0, Week 1, Week 2, Week 3, and Week 4
• The above blood pressure results are an example of experimental units or subjects; factors, features or variable and treatment/factor/intervention.
• Components:
  • Hypothesis:
    • is drug A is different from drug B or their are no difference?
  • Control:
    • Two or more treatments should be compared is to Ctr, drug A drug B
  • Randomization:
    • The experimental units should be randomly divided into groups can matched age/gender is done
  • Replication:
    • An efficient amount of experimental units is 6-biological replicates
  • Blocking:
    • Results can depend on Compare results within blocks is potential confounder

Correct experimental design:

• Hypothesis:
  • Drug A is different from drug B.
• Control:
  • Two or more treatments should be compared: Group Ctr, drug A and drug B.
• Randomization:
  • The experimental units divide randomly into groups to avoid selection bias.
  • Including matched the age and gender
• Replication:
  • A sufficient number of experimental units: Six biological replicates

Types of Experiments

• Completely Randomized Experiment:
  • Each unit randomly assignd to different groups to revive treatments
  • The unit in the some group will receive the same treatment
  • Compares Results of each treatment betwwen groups
  • Advantage: simple to implement and reduces the risk of contamination between experimental conditions.
  • Limitations: Requires more participants, possible variability between groups can confound results and individual differences.
• Randomized blocking design:
  • The whole cohort divides based on the blocker into "blocks".
  • The units in each block compares internally. and takes results of recording
• Matched paired design
  • individual and variation experiment with effect
  • Matched pair in two treatments we compares in paired unit
• In-randomized experiment, individual variation has an effect
• In Matched paired, we compare the two treatments in the same unit {paired}
• Grace period in drug studies, are to be considered
• Advantages requires fewer participants, and controls for individual differs
• Limitations potential of carryover effect treatment from previous affecting next.
• Factorial design
  • Involves studying the effects of two or more independent variables (factors) simultaneously.
  • Multiple levels per factor that can multiple combinations
  • Factor 1 for drugs Factor A and B
  • Factor 2 for Low and High Doses
  • Advsantages for individual and combined effects from multiple factors and efficient for participant usage
  • Limitation is more complex
• Quasi experimental design
• A non-random unites assignment compared to groups
• a pre-existing and naturally groups for the non-independent varibales
• Useful not non-possibility of randomness

Types of Studies

• Ecological studies
• group of popualtion used for feature or Phenonom on for Sociocoty systems.
• Data results and the other includes are at popaulaitons level
• Used with highpthersis use with free researches
• Examples like different rate or disease with different county
• Also Generates highpthersis by questions/generatoin for future Investigation and Comparing population that are over-timed .
• Cross-Section study
  • Is to asses a population where exmaiatn to a variables reailsihnsiop
  • repersnttaive population should study undertaken should
• Is quick
• Is Inexpnsive
• Cant do caausilty
• Examples COVID-19 in the UK in 2024 in the study
• Case report and Case Series, with a Detailed observaoty indviuaks reports and the serie reports multiple

Case-Control studies

• Compare inidvaiuas with coinditona of case and used for idnfeitcaotp of risk and predict like case factors
• the factors like high risks used the conditonal reports
• Cohort Study
  • Following group for rsk that has determineto find reuslats
  • EXAPMPLE stduys that follows from to non compare with over lung c acner time by two periods years

Experiments

• Testing for efficiency of new vaccine again COVID-19 infection -Randomized block study: - Antibody tiers by Continuous Data: and viral tiers by contnousu Data. -The system should track of wasout by periods and should ensures the labeling is porpoer - Full Factory Design -Antibody titres by Continous data and Virul Teris used by continous by data to a timing effects