Computer generated transcript

Warning!
The following transcript was generated automatically from the content and has not been checked or corrected manually.

ICA3:AMedEdLecture Abhisekh ChatterjeeDisclaimer This is not official advice from the BSc course. Please do not take it as such. The slides and recording for this talk willbe availableafterwards. Any specific questions, please email me: ac2420@ic.ac.uk. Please do not send me your ICA 3 to look at.Contents • What is ICA 3? • What’s the point of this talk? • What should be your approach? • Data analysis and statistics • Writing • Introduction, Abstract and Lay summary • Useful links, resources and tipsWhatisICA3? A“data analysis”task. You are given some raw data (you may have already collected this yourself), given a brief idea about why these experiments were done. You then analyse the data, figure out why these experiments were done, produce figures and write a 1500 word document (essentially a minipaper) about these experiments.What’ sthepointofthistalk? • Where do most students strugglewith ICA 3? • Thisis nota statistics lecture–well beyond the scope. • now.ll cover it brieflybut it isfar moreimportantfor the finalICA than itis • and all the other ICAstoo.ally good stepping stone todo really well in thisICA • The most difficult aspects of this ICA are (generally) from easiest tohardest: • Statisticsand makinggraphs (stay tunedfor mytips)ier) • Crammingeverythingin 1500words (canseemimpossible) • Figuring out WHYON EARTH THEYDID THE EXPERIMENTS • Criticalsynthesis/criticalanalysis/criticalsummarization(?) – whatis this?Quickpoints • Some people have already done their ICA 3. • You already have your data. • Undoubtedly, you have been discussing your data and its analysis. • You probably already know what statistical tests you need to do, and what the results are. • What now?StructureandApproach Lay summary (500) Don’tworry – papers are not written Scientific abstract (300) in thisformat. constructinghypotheses. Plan andd Results compendium write mostof your introduction.Then (1500): moveon to the trickypart – Methods “hypotheses”section of thee introduction,discussion and Results conclusion. Discussion Abstractand lay summary absolutely, in no uncertain terms, Conclusion comelast.Statistics eventually understand why this is often outsourced to experts.ata analysis, you’ll My top tips: • Understand the type of data you have (categorical, continuous, ordinal, etc.) • Assess normality (Shapiro-Wilks or KS) – this may be optional in your course. • Conduct your statistical test(s) of choice. This will probably be a lot of independent T-tests (paired T tests for the same group at 2 different points in time – you probably will not use this). • There are tests to be used for parametric and non-parametric data respectively. • The Bonferroni multiple test correction (may be optional!) • List absolutely everything that you did and in each figure (more on this later) give the * corresponding to p-values.StatisticsPt.2 • Undoubtedly,youprobablyalready know whattests youneed to use/haveused. • The final slide of this talk has some really helpfulresourcesif youwant to learn more,which maycomein veryhandyfor the final ICA. • Software • Itis not worth yourtimeto go andlearn an entirelynew programminglanguagejustto make6 simple figures. • IF youhave the time,energy andinterest,R (usingthe Tidyversecollectionof packages)isa goodplace to start and willset youupnicelyforthe rest of youracademiccareer. • PRISM’smore updatedversioncanbedownloadedto yourlaptopand usedwith a free trial. • Alternatively,availableforfree onImperialcomputers. • PRISMmakesabsolutelybeautifulgraphs. • Excelisquick,easy and (generally)painless. • Use whateverisrecommendedor easiestto get yourhands on. • PRISM is generally really excellent as it has statistical tests embeddedintothe software itself, and thenewer versions will automaticallydo thep-valuebars for you.Figures Export your graphs as PNGs or JPEGs and import them into a PPT slide. Then construct therest of your figureand when you’re done export each individual slide as a PNG with high DPI tomaintain quality. Important points: • The figuretitle should be a one line conclusion about the experiment. People should be able to look at it and get the general idea about what the main takeaway is from that figure. • Your figurelegend should be a brief methodology, giving details of the names of the eg. Mice or cells that wereused, including timings and dosages. • Give the gender of the animals (if applicable), a sample size(n = ?),how the data is used and correction (if applicable), and finally the * corresponding top-values and any abbreviations (HFD: high-fat diet).Writing Now for the difficult part! a) Figuring out your hypotheses. b) How do you do this“critical synthesis”thing????? c) Actually writing.Figuringoutyourhypothesesandnarrative • Unfortunately, I can’t give a huge amount of specific advice here! • Useyour intuition. • Work backwards. • Look at other key papers in the field. • Remember – you are tellinga story. Make sure there is a cohesive narrative that can be followed in your writing,starting from your introduction.CriticalSynthesis • One of the most nebulous concepts in the BSc, but easily the most important if you want to do well. • What separates good writers from excellent writers. • Generally the mostsought after skill from academic doctors. Anyone can write code, collect and/oranalyse data – but can you analyse it appropriately?CriticalSynthesisPt.2 “Aet al. did this. B et al. did this and foundthis. Conversely,C. et al. did somethingdifferentandfound this.” Unlessthere isa key paper youare focusingon,orthere aren’t that manyinthe fieldofstudy, namedroppingauthorsandlistingthe results ofeach individualpaper isnot critical synthesis.Instead, methodologyvaried,orhas the technologyimproved?Are the results surprisingornot?Are thereas the confoundingvariables?Are there limitationsto yourwork,or the otherpapers? Just because onevariable haschanged,doesn’t mean that others will! Havea read of Nature paper discussions.A gooddiscussionwill answerseveralquestions: - Whatdoes ourwork show? - How does ourwork fitwith the rest of the literature? - Whyare ourresults significant? - Whatworkcan be doneinthe futurebased onthis work/expand uponit? - Whatare the limitationsofourstudy?Writingtoptips • A really good criticalsynthesis is an understanding of the literature and the contextof your work inside it, creating connectionsand bridges and finding differences in places youleast expect them. • More often than not,youcansummarise a lotof literature that is background byjust makinga broad conclusionabout themand referencing them. • For example:“Previous work has demonstratedX to be a vital componentof health-related interventions [REF]. ” • Youare notwriting a novel. Keep your sentences short,succinctand to the point. • Linking words help maintain the flow of your writing.TheIntroduction • Introduce your story. • Helpful to gather whatever references you can ahead of time/what you are currently doing. • You will have to condense a fair amount of literature. • Try and find the pertinent question/gap in the literatureand base everything else off this.TheAbstractandLaysummary • Write your scientific abstractfirst so youcan translate it into“normal speak”for thelay abstract. • Givepertinent results (mayor maynotinclude values)– in plain English – witha pvalue. • Eg GroupA exhibited a greater reduction in fat free mass (p<0.05)than GroupB mice. • Lay summary– this is actually really easy. Lookup a nature short communication– itshould be written in thisstyle. • Imagine someone non-scientificis reading yourlay summary (maybe have them read it as well) • Throw somehumour in there!Resources Statistics“cheat sheet”: https://www.statstutor.ac.uk/resources/uploaded/tutorsquickguidetosta tistics.pdf Nature shortcommunications(thisis a great example butyou canread any that aren’tan editorial reply): https://www.nature.com/articles/4441022a StatisticsYouTube videos: https://youtu.be/ijeEYFnS2v4?si=haGp11CpLzfo_t3 This YouTube channelhas some excellent videos.THANKYOU FOR COMING! PLEASE FILL IN THE FEEDBACK FORM!