paper 3 Flashcards
(37 cards)
1
Q
what statistical tests should be done if data is in 2 groups?
A
parametric: t-test
non-parametric: Mann-Whitney U test
2
Q
what statistical tests should be done if data is categorical not numerical?
A
compare 2 variables: Chi-squared
test proportions: Z-test (binomial)
2
Q
what statistical tests should be done if data is paired?
A
parametric: paired t-test
non-parametric: Wilcoxon signed-rank test
3
Q
what statistical tests should be done if data is more than 2 groups?
A
- parametric: ANOVA
- non-parametric: Kruskal-Wallis test
4
Q
what statistical tests should be done if data has multiple variables you are seeking the relationship between?
A
- parametric: Pearson correlation regression
- non-parametric: Spearman correlation
5
Q
what are the key points to include for Q1 central aims?
A
- overall aim, identifies what paper set out to do, and what it proves: reword title
- specific aims that recap each experiment
- summarise main findings
- highlight IMPORTANCE, what is unique about the study
- did they use interesting techniques?
- place aim in context of the field
- what they tested, which led to what they found
6
Q
what are the key points to include for Q1 key figure?
A
- the figure which recaps the central aim + best proves results they set out to find
- describe what they did, what they found
- explain techniques + validation of those techniques
- why does data in the figure support the central aims
- can the results from the figure demonstrate causality?
7
Q
when describing the experiment/technique, include what?
A
- specificity!!
- chronologically explain experiment
- why was experiment designed in the way it was
- data analysis + post-hoc analysis
- why was it conducted?
- how should results be interpreted?
8
Q
advantages + disadvantages of Cre transgenics
A
- A: highly specific expression under a tissue-specific promoter
- D: every cell with gene expressed can be activated
- sometimes lack of full transfection
- numerous controls + postmorterm to validate expression specificity
- insertions of seqs can disrupt endogenous genes, disrupt cognitive/emotional circuitry
9
Q
advantages + disadvantages of viral injection transgenics?
A
- A: spatial specificity based on site of injection
- temporal specificity
- target specific cell types if combined with cre recombinase lines
- D: success depends on levels of transduction in target neurons, variability common
- off-target effects due to non-uniformity of light scattering
- potential damage + inflammation
10
Q
advantages + disadvantages of behavioural studies?
A
- D: difficult to interpret in some rodents
- genetic modification can alter normal behaviour, or cause more general deficits
11
Q
advantages + disadvantages of electrophysiology?
A
- A: combine brain + muscular activity measurements
- monitor brain states
- quantifiable
- good temporal resolution
- D: poor spatial resolution
- passive recordings don’t confer causal information
12
Q
advantages + disadvantages of viral GFP labelling
A
- A: labels whole cell (axons + dendrites)
- can be used in living preparation, which can be exposed to multiple techniques
- target injections to region of interest
- D: injections damaging + inflammatory
- retrovirus can infect other cells
- BrdU is more efficient in labelling dividing cells
- variable amount of virus injected (precision + accuracy)
- GFP not always fluorescent, may require IHC
13
Q
advantages + disadvantages of 2p microscopy
A
- A: allow imaging at depth 10x confocal bc longer wavelength
- less photobleaching + phototoxicity
- very good spatial resolution bc narrow plane of focus
- live imaging, non-damaging
- different colours
- within subject comparisons
- D: stress caused to animal with multiple recording devices
- tracers can interact with each other
14
Q
advantages + disadvantages of optogenetics
A
- A: can manipulate neural activity in freely moving animals (close to physiological conditions)
- reversible
- high temporal + spatial specificity
- animal serves as its own control
- activation is more selective than direct electrode activation
- D: need to check for lesions at surgical site, correcti fibre optic positioning
- light scattering –> not all cells equally activated/inhibited
- need to implant the optical fibre
15
Q
advantages + disadvantages of chemogenetics
A
- A: longer timeframe, better for behavioural studies
- fully reversible
- high temporal + spatial res
- non-invasive compared to optogeneics
- animal can be its own control
- D: need strict controls
- need to target the entire population of cells expressing the gene
- neurons can respond variably to GPCR action, need to do electrophys in vitro to see how neurons respons
16
Q
what is chemogenetics?
A
- uses genetically engineered receptors to modualte neural activity/pathways using specific small molecule drugs
- uses chemical ligands rather than opsin channels
- before this, TMS and DBS were used to study relationship between neuronal activity and behaviour
17
Q
what are the most common GPCRs used in chemogenetics?
A
- DREADDs, get activated solely by drug of interest
18
Q
why is chemogenetics now favoured over optogenetics?
A
- C does not require expensive light equipment
- resolution in O declines due to light scattering + illuminance declining over distance -> leads to lower spatial resolution
- C represents a pharmacological avenue
19
Q
what to consider when critiquing data
A
- are methods appropriate?
- is actual data good?
- have multiple techniques been used?
- are statistics valid?
- have results been overinterpreted?
- are controls good?
- correlation vs causation, necessary vs sufficient?
- propose methods to improve weaknesses
20
Q
what to look for in stats?
A
- what do error bars represent?
- check for independent means, not subsamples
- is data normally disitributed (if n > 10, central limit theorem kicks in)
- spread of data
- is test appropriate? same variances?
- what about the power of the study?
21
Q
how to increase power of statistical test?
A
- increase sample size
- reduce measurement error
- increase effect size
- increase significant level
- use one-tailed instead of two-tailed
22
Q
what is effect size and how can you increase it?
A
effect size = a measure of strength of relationship between 2 variables in a population i.e. correlation, regression coefficient, mean difference, risk
increase it: reduce standard deviation of it, removing irrelevant particiapnts, optimising experimental designs + measures
works well when massive sample sizes are not feasible, for example in psychologically interesting phenomena
23
Q
what is a big or small effect size?
A
- small: d < 0.5
- medium: 0.5 < d < 0.8
- large: 0.8 < d < 1.3
24
how to design a good experiment?
* CLARITY OF DESIGN
* show causation
* justify well how it would advance the field
* propose hypothesis
* produce experimental design
* add experimental details + controls
* recognise expected outcomes
* how to interpret proposed findings
25
what are some lines of enquiry for a follow up expt?
* change technique to validate same results
* change animal model to show translation of results
* relate results to humans
* relate results to disease/disease model
* provide more detail to exisiting results
* explore downstream targets
* look at same thing in different brain area
* increase specificity of cell types and locations
* overexpress something that was KD'd to prove same thing in opposite way, shows effects of KD are specific and not RNAi artefacts
26
what are some general strengths of a paper?
* use of multiple techniques
* demonstrating causality
* ability to manipulate activity, ideally in both directions
* confirmation of results in multiple species
* adequate control populations
* use of state of the art tech
27
what are some general weaknesses of studies?
* use of limited number of techs
* no linking results to wider context/global view
* no consideration of related fields
* missing varied expts
* lack of info about specific circuitry/cell types
* are histology images randomly chosen or representative?
* how good is stats?
* things being inferred e.g. from gene expression rather than directly assayed
* tests being carried out only within conditions not between conditions
28
how to write a lay summary
* what do authors conclude, not what your take is
* broad concept, relate concept to paper
* what does this study show + conclusions
* what did they do + result
* overall conclusions
* what does it mean for future/wider field?
29
what is the point of a post-hoc test?
if ANOVA is significant, need post-hoc to compare means
* most post-hocs control for multiple comparisons, easiest is Bonferroni correction (p/number of means) e.g. 0.05/3
* used when several stats tests are performed simultaneously
30
whats the difference between what SD and SEM show?
* SD: descriptive + inferential, tells you nothing about differences only spread + variability
* SEM: gives comparisons differences between means, how precise are the estimates of means? how close is sample mean to population mean? if no overlap between groups, need stats test but probs significant
31
how can effect size be manipulated in bad faith?
* if you have a large enough sample size, error shrinks and minute differences become statistically significant: but is the difference useful?
* stats tests only work bc they're dependent on the probability of random events, if you keep replicating an expt to increase n until you have significant results, you violate assumptions
32
what to do w strengths + weaknesses Q
* 5 strong points
* more replicates BECAUSE increases power
* relate technical limitations to conclusions drawn in the paper
* reflect how good the paper actually was
* can recap answers from previous questions
33
what to include in the abstract?
* introduce subject area correctly + concisely
* clearly state experimental aims
* state key findings
* give clear impression of overall results/conclusions for a general readership
* appropriate conclusion sentence outlining broader implications of finding
34
how does one decide whether an appropriate number of participants have been recruited to a study?
* specific challenges of integrative physiology: cost, compliance, adverse effects
* see if formal power calculations are supplied by paper
35
difference between post hoc test and post hoc with multiple comparisons?
* post hoc = to identify exactly which groups differ from another
* post hoc w MC = controls family-wise error rate to control rate of type I (positive) errors at the stated p (alpha)
36
how to drive cell-specific gene overexpression with Cre-LoxP?
* place cre under tissue specific promoter
* place lox-stop-lox cassette upstream of ectopic gene, its removal by recombinase would drive overexpresion
