Science Skills Boot Camp

Workshop 2: “Science is a way of thinking Asking Scientific Questions much more than it is a body of knowledge.” Keren Witkin, PhD Carl Sagan

www.training.nih.gov

Choose research question The scientific Where does this fit into the method Research scientific method: background

Prove that your Construct hypothesis hypothesis is true

Perform investigation

Analyze results and draw conclusions

Today’s investigation: Class data collection 1: (PTC) Taste the PTC paper and pick one description:  This tastes like paper  This tastes a little bitter The human TAS2R38 encodes  This tastes horribly bitter taste receptor that detects PTC

http://learn.genetics.utah.edu/content/begin/traits/ptc/ Please dispose of used PTC paper into the ziploc.

1 What makes a good research question?

Step 1:  Testable Choosing a  Not too broad, but not too narrow research  Fits appropriately into the existing information question  Realistic in terms of resources  Contributes to the field no matter what you find

Some possible questions:

 How does the sense of taste work?  Do factors other than genetics affect the ability Step 2: to taste PTC? Researching  How did PTC sensitivity evolve? background  Are supertasters cuter than non-tasters?  Are there gender differences in PTC sensitivity?  Do PTC tasters have a specific DNA ?

Finding background information Review articles are a great place to start!

 Provide an overview of the field  Primary literature  Often written by experts in the field  Review articles  Summarize many primary papers  Textbooks  Often contain useful diagrams  Your colleagues  Internet resources

Webster et al. Journal of Cell Science. 2009. 122: 1477-1486

2 Primary literature All about PubMed  This is where the actual data is!  Free database of biomedical and life sciences  Describes the experiments in detail literature  Introduction  Maintained by the National Library of Medicine  Methods  Contains over 20 million citations  Results  http://www.ncbi.nlm.nih.gov/pubmed/  Figures  Tables

 Discussion

 References

3 If there’s no link from PubMed, the NIH library may still have it

 Go to the NCI Frederick Scientific Library webpage: http:// www-library.ncifcrf.gov/default.aspx  Click on Online Journals (under Resources)

If there’s no Online access. . . Where else can you find scientific papers?  Go to the library (Building 549)  EMBASE: Biomedical and pharmacological  Document delivery (Under services on library database Website)  SCOPUS: Broad coverage of scientific, technical, medical, and social science literature, including arts and humanities  Web of Science: Coverage of ~12,000 journals and conference proceedings. Not limited to biomedical sciences.

Also on the Internet: Bioinformatics tools  GenBank

 Free annotated collection of DNA sequences, run by NCBI (NIH)

 http://www.ncbi.nlm.nih.gov/genbank/

 Search by gene name to get DNA or protein sequence

 “Blast” a DNA or protein sequence to find matches in the database

4 More bioinformatics tools Step 3:  Sequence alignment programs Constructing a  Structure prediction programs hypothesis  Gene expression and regulation databases  Organism-specific databases

 Pathway analysis Hypothesis:  Promoter/SNP prediction Tentative statement predicting the outcome of an experiment And many others! If . . ., then . . .

What we know about taste: What we know about taste: Taste is controlled by receptors in the tongue Taste perception involves additional factors

 Smell  Temperature  Visual cues  Texture

http://health.howstuffworks.com

What we know about taste: Studies show that smelling mint while eating Step 4: makes it difficult to identify food by taste. Perform investigation

Hypothesis: If you smell mint while tasting PTC paper, then you will be less sensitive to the bitter Planning Doing taste.

5 Planning experiments Planning experiments For each experiment: For each experiment:  Define your objective  To determine whether strong smells decrease sensitivity  Define your objective to PTC paper  Plan your general strategy  Plan your general strategy  Decide on experimental details  Participants will taste PTC paper and report on the taste. They will then smell mint while re-tasting PTC paper and report on the taste.  Decide on experimental details  How many participants? What kind of mint? How will they rate PTC taste? What will be the controls?

Common controls What are some features of a good

experiment?  Positive controls Show that everything is working well, and that your  Has a clear purpose conditions are able to achieve a positive result, even if  Answers one question definitively your samples turn out all negative  Has appropriate controls  Negative controls  Has limited variables Show the base-line background in your experiment,  Has a large enough sample size using known samples that should produce a negative result  Uses available reagents and equipment  Can be repeated by you and others Sample - + Sample

Our experiment today: Each subject will taste PTC paper and  What are the controls in this report whether there is no taste, weak bitter taste, or strong bitter taste experiment?

 What are the variables? Experimental Group Control Group Half of the subjects will re- Half of the subjects will re- taste PTC paper while taste PTC paper while  Are there other controls that we’re smelling mint and will report smelling parsley and will missing? whether there is no taste, report whether there is no weak bitter taste or strong taste, weak bitter taste or bitter taste strong bitter taste

6 Tips for good experiments Step 4:  Be prepared Perform  Be organized investigation  Be meticulous

 Work deliberately and carefully

 Follow protocol closely

 Note any deviations from protocol  Minimize bias Planning Doing  Document everything

Learning a new technique Making or acquiring reagents

 If ordering reagents, do it as early as possible  Find a protocol  Research how each reagent should be used and  Read it carefully stored  Consult with your mentor or other experts  If making up solutions  Make or acquire reagents ahead of time  Make sure you know what solvent to use  Learn how to use required equipment  Brush up on molarity and serial dilutions, if  Do a “dry run” necessary  Allow plenty of time for the first run  Check and re-check all calculations

Common mistakes What if your experiment doesn’t work??

 Doing huge experiments with too many samples  Not thinking carefully about your controls before you start

 Waiting until the last minute before tracking down reagents

 Forgetting to grow up the cells you need ahead Frustration of time Time (days)

7 When your experiment doesn’t work: Class data collection- Part 2 Troubleshooting 1. Taste the PTC paper while smelling the mint  Identify possible sources of error or parsley and pick one description:  Reread protocol  This tastes like paper  Check calculations  This tastes a little bitter  Consider whether reagents or equipment might be  This tastes horribly bitter suspect

 Think about repeating experiment as is 2. Dispose of all waste into the ziploc  Consult with mentor  Consider whether your hypothesis might be 3. Class data will be collected and recorded on flawed the board  Don’t get frustrated!

• How many people changed their responses after smelling the mint? Step 5: Analyzing • How many people changed their responses in Results and the control group? Drawing Conclusions • Can we draw any conclusions? • Does our data suggest future experiments?

• How do we want to present this data?

Parsley Mint Raw data: Recorded in your notebook Processed Before After Before After data: No taste 12 12 13 13 Weakly Subject Mint or Before After 27 25 23 24 parsley bitter Other types Strongly 11 13 14 13 1 M No No of raw data: bitter 2 M Strong Strong Gels 30 PTC tasting results 25 Blots Before 3 M Strong Weak 20 Photographs After 15 4 M Weak Weak Observations 5 P No No 10 Number of subjects Numbersubjects of 5

6 P Weak Strong 0 no taste no taste weakly weakly strongly strongly 7 P Weak Weak (control) (mint) bitter bitter bitter bitter (control) (mint) (control) (mint)

8 Another example: Is the data consistent with your hypothesis? If not, then start again at step 3!

extracts Construct hypothesis Esp1-FLAG Marker - 1 2 3 4 200 kD 116 kD Perform investigation Raw data Processed data

Analyze results and draw conclusions

Data vs interpretation Question: Data: 48/50 study participants reported no How could we design an change in PTC taste sensitivity after smelling mint, compared to 49/50 in the control group. experiment to better test the broader conclusion that Interpretations: smell does not affect taste? • Smell does not affect taste • Smelling mint does not affect taste • The smell of mint does not affect PTC taste sensitivity

An example of a difficult ethical question Does your data mean what you think it Predicted data: means? A curve similar to this one

 Is it statistically significant? Actual data:  Are you doing the right statistical analysis? All of these data points

 Do you have a large enough sample size or Possible source of error: enough repetitions? Power supply fluctuations  Are there alternative explanations?  Are there confounding factors? The ethical question: Can you discard the two filled squares as outliers?

Adapted from “On Being a Scientist: Responsible Conduct in Research,” 2nd ed., National Academy Press, 1995.

9 Pitfalls in data analysis Scientific misconduct

 Falsifying data  Only considering specific data points Can happen accidentally when you “process”  Over-interpretation of data data  Ignoring confounding factors  Fabricating data  Using too small of a sample size This is always wrong!  Plagiarism Includes using other’s IDEAS as well as WORDS

Same basic scientific method no matter Mathematical what kind of research you do tools • Basic research Computational Choose Experiments question analyses • Clinical research Research • Translational research background Construct • Social and behavioral research hypothesis Perform Input • Epidemiology investigation Results Analyze results • Computational research and draw conclusions • Mathematical modeling Does the data support my hypothesis??

Special considerations for Special considerations for clinical epidemiologists research  Are you surveying an appropriate population?  Bioethics  Do you have enough study participants?  Professionalism  Are you using the appropriate analytical tools?  Confidentiality  Have you considered potential alternative  Possibility of health risks for investigator explanations?  Critical to have appropriate study design and  Confounding factors meticulous technique

 Bias  Institutional Review Board (IRB)

 Chance  Data Safety Monitoring Board

 Reverse causality  Double-blind studies

10 More on bioethics: The 7 major ethical Special considerations when using principles that guide clinical research animals in research  Oversight by the NIH Office of Laboratory Animal  Social and clinical value Welfare  Scientific validity  Must have an Animal Care and Use Protocol Each study  Fair subject selection  Ensuring humane and responsible use must have:  Favorable risk-benefit ratio  Carefully designed experiments  Independent review  Minimizing the number of animals used  Avoiding/minimizing pain and stress  Informed consent  Appropriate housing conditions  Respect for potential and  Appropriate sedation or anesthesia enrolled subjects  Veterinary care, when necessary

Ethical considerations for all kinds of Last but not least . . . scientists • Honesty

• Objectivity Communicate your • Integrity results • Carefulness • Respect for intellectual property • Responsible publication • Respect for colleagues

11