2013 10-16-sbc3610-research methcomm

43
Research Methods & Comms. [email protected]

description

Research methods & comms Some info about careers, about writing, introduction to R practicals (regular expressions, functions, loops), experimental design. @ Queen Mary U London.

Transcript of 2013 10-16-sbc3610-research methcomm

Page 1: 2013 10-16-sbc3610-research methcomm

Research Methods & Comms.

[email protected]

Page 2: 2013 10-16-sbc3610-research methcomm

© Alex Wild & others

Page 3: 2013 10-16-sbc3610-research methcomm
Page 4: 2013 10-16-sbc3610-research methcomm

© National Geographic

Atta leaf-cutter ants

Page 5: 2013 10-16-sbc3610-research methcomm

© National Geographic

Atta leaf-cutter ants

Page 6: 2013 10-16-sbc3610-research methcomm

© National Geographic

Atta leaf-cutter ants

Page 7: 2013 10-16-sbc3610-research methcomm
Page 8: 2013 10-16-sbc3610-research methcomm

Oecophylla Weaver ants

© ameisenforum.de

Page 9: 2013 10-16-sbc3610-research methcomm

© ameisenforum.de

Fourmis tisserandes

Page 10: 2013 10-16-sbc3610-research methcomm

© ameisenforum.de

Oecophylla Weaver ants

Page 11: 2013 10-16-sbc3610-research methcomm

© forestryimages.org© wynnie@flickr

Page 12: 2013 10-16-sbc3610-research methcomm

Tofilski et al 2008

Forelius pusillus

Page 13: 2013 10-16-sbc3610-research methcomm

Tofilski et al 2008

Forelius pusillus hides the nest entrance at night

Page 14: 2013 10-16-sbc3610-research methcomm

Tofilski et al 2008

Forelius pusillus hides the nest entrance at night

Page 15: 2013 10-16-sbc3610-research methcomm

Tofilski et al 2008

Forelius pusillus hides the nest entrance at night

Page 16: 2013 10-16-sbc3610-research methcomm

Tofilski et al 2008

Forelius pusillus hides the nest entrance at night

Page 17: 2013 10-16-sbc3610-research methcomm

Avant

Workers staying outside die« preventive self-sacrifice »

Tofilski et al 2008

Forelius pusillus hides the nest entrance at night

Page 18: 2013 10-16-sbc3610-research methcomm

Dorylus driver ants: ants with no home

© BBC

Page 19: 2013 10-16-sbc3610-research methcomm

Animal biomass (Brazilian rainforest)

from Fittkau & Klinge 1973

Other insects AmphibiansReptiles

Birds

Mammals

!Earthworms

!!

Spiders

Soil fauna excluding earthworms,

ants & termites

Ants & termites

Page 20: 2013 10-16-sbc3610-research methcomm
Page 21: 2013 10-16-sbc3610-research methcomm

Big data is invading biology

Page 22: 2013 10-16-sbc3610-research methcomm

This changes everything.454

Illumina Solid...

Any lab can sequence anything!

Page 23: 2013 10-16-sbc3610-research methcomm

Big data is invading biology• Genomics

• Biodiversity assessments

• Stool microbiome sequencing

• Personalized medicine

• Cancer genomics

• Sensor networks - e.g tracking microclimates

• Aerial surveys (Drones) - e.g. crop productivity; rainforest cover

• Camera traps

Page 24: 2013 10-16-sbc3610-research methcomm
Page 25: 2013 10-16-sbc3610-research methcomm

Learning to deal with big data takes time

• New Master’s Programs @ QM:

• Bioinformatics (for biologists)

• Ecological & Evolutionary Genomics (or Biodiversity Informatics)

• Our 6 hours of practicals.

Page 26: 2013 10-16-sbc3610-research methcomm

Practicals

• Aim: get relevant data handling skills • Doing things by hand: slow, error-prone, often impossible. • Automate!

• Basic programming • in R • no stats!

Page 27: 2013 10-16-sbc3610-research methcomm

Practicals: format• Groups - ok? • 3h practical this week

• data accessing/subsetting • search/replace • regular expressions

• 3h in two weeks • 2h practical

• functions • loops

• 1h exam (last hour of practical)

Page 28: 2013 10-16-sbc3610-research methcomm

http://tryr.codeschool.com

Page 29: 2013 10-16-sbc3610-research methcomm

Regular expressions: Text search on steroids.

Regular expression FindsDavid David

Dav(e|id) David, DaveDav(e|id|ide|o) David, Dave, Davide, Davo

At{1,2}enborough Attenborough, Atenborough

Atte[nm]borough Attenborough, Attemborough

At{1,2}[ei][nm]bo{0,1}ro(ugh){0,1} Atimbro, attenbrough, etc.

Easy counting, replacing all with “Sir David Attenborough”

Page 30: 2013 10-16-sbc3610-research methcomm

Regular expressions

• Google “Regular expression cheat sheet” • ?regexp

Synonymous with\d [0-9]

[A-z] [A-z], ie [A-Za-z]

\s whitespace

. any single character

.+ one to many of anything

b* between 0 and infinity letter ‘b’

[^abc] any character other than a, b or c.

\( (

[:punct:] any of these: ! " # $ % & ' ( ) * + , - . / : ; < = > ? @ [ \ ] ^ _ ` { |

Page 31: 2013 10-16-sbc3610-research methcomm

Functions• R has many. e.g.: plot(), t.test()

• Making your own: tree_age_estimate <- function(diameter, species) { [...do the magic... # maybe something like: growth.rate <- growth.rates[ species ] age.estimate <- diameter / growth.rate ...]! return(age.estimate)}> tree_age_estimate(25, “White Oak”)+ 66> tree_age_estimate(60, “Carya ovata”)+ 190

Page 32: 2013 10-16-sbc3610-research methcomm

“for” Loop

> possible_colours <- c('blue', 'cyan', 'sky-blue', 'navy blue', 'steel blue', 'royal blue', 'slate blue', 'light blue', 'dark blue', 'prussian blue', 'indigo', 'baby blue', 'electric blue')!> possible_colours [1] "blue" "cyan" "sky-blue" "navy blue" [5] "steel blue" "royal blue" "slate blue" "light blue" [9] "dark blue" "prussian blue" "indigo" "baby blue" [13] "electric blue"!> for (colour in possible_colours) {+ print(paste("The sky is oh so, so", colour))+ }![1] "The sky is so, oh so blue"[1] "The sky is so, oh so cyan"[1] "The sky is so, oh so sky-blue"[1] "The sky is so, oh so navy blue"[1] "The sky is so, oh so steel blue"[1] "The sky is so, oh so royal blue"[1] "The sky is so, oh so slate blue"[1] "The sky is so, oh so light blue"[1] "The sky is so, oh so dark blue"[1] "The sky is so, oh so prussian blue"[1] "The sky is so, oh so indigo"[1] "The sky is so, oh so baby blue"[1] "The sky is so, oh so electric blue"

Page 33: 2013 10-16-sbc3610-research methcomm

xkcd

Page 34: 2013 10-16-sbc3610-research methcomm

1.23 (?)

Page 35: 2013 10-16-sbc3610-research methcomm

More career stuff

• Internships?

• What does PhD mean?

• Basic CV rules

Page 36: 2013 10-16-sbc3610-research methcomm

Writing

• an essay

• a cover letter

• a reference letter

• (a “new scientist article”)

• a dissertation

• (an abstract)

Page 37: 2013 10-16-sbc3610-research methcomm

QMUL marking schemeMarking Criteria and Mark Scheme for Essay-style Questions

Levels 5 - 6 Level 6

Evidence of Comprehension Breadth and Depth of Knowledge Irrelevant Material and Errors Synthesis & Balance Originality & Innovation

A+ Outstanding. Deep insightOutstanding. As much as could be expected

Absent or minimal Evidence of critical analysis Original ideas and insight

AClear understanding. Shrewd and appropriate

Extensive. Almost as much as could be expected

Minimal or absent Astute selection and juxtaposition Some evidence of creative interpretation

A- Tending to description rather than interpretation

Extensive Minimal Appropriate selection and combination Some

A--Sufficient to marshal a well-organised, direct response

Most key points but not extensivePerhaps some minor errors and tangential material

Inappropriate balance, partial synthesis Limited

BSufficient to marshal an organised, direct response

Not all key points but comprehensive and accurate

Some minor errors and tangential material Inappropriate balance, partial synthesis Limited

CNot a direct response but sufficient for a logical presentation.

Several omissions but some key points Some errors, tangential material Minimal Minimal

D,EPoor comprehension, muddled organisation

Major omissions. No key points. A few basic facts

Major factual errors. Frequently irrelevant None None

F+ Almost none One or two very minor points correct Extensively irrelevant or wrong None None

F NoneOne or two very minor points just about correct

Extensively irrelevant or wrong None None

F- None No evidence of being better if longer Almost all irrelevant or wrong None None

0 Nothing written Nothing written Nothing written None None

All Levels(Desirable in other years)

Notes:

x In order to qualify for an "A-grade" the work must meet most of the indicated criteria.

x� Grade to % conversion: A+ = 100; A = 92; A- = 83; A-- = 74; B+ = 68; B = 65; B- = 63; C+ = 58; C = 55; C- = 53; D+ = 49; D = 48; D- = 47; E+ = 44; E = 43; E- = 42; F++ = 39; F+ = 37; F = 27; F- = 17; 0 = 0

JV / TLC (23/01/2013)

Page 38: 2013 10-16-sbc3610-research methcomm

Important for all: StructureClear overall structure?

Separate intro starts from general points. announces the structure (paragraphs or major sections).

One paragraph per idea/point. Clear structure within each paragraph.

If includes a list: “three lines of evidence suggest that X. First, ...., Second, ... Finally....”

Clarity of each sentence. No unnecessary words!

Try to make smooth transitions

Page 39: 2013 10-16-sbc3610-research methcomm

More writing tips. • No ping-ponging! • No unnecessary ideas. • Eliminate unnecessary words.

• “We have performed X” --> “We did X” • shorter is better

• Put MS Word in “strict grammar” • Eliminate jargon.

• write for the “general smart scientists” with little domain specific knowledge.

Page 40: 2013 10-16-sbc3610-research methcomm

Writing

• an essay

• a cover letter

• a reference letter

• (a “new scientist article”)

• a dissertation

• (an abstract)

Page 41: 2013 10-16-sbc3610-research methcomm

Rest of our time together

Experimental design

(Reproducible research)

Page 42: 2013 10-16-sbc3610-research methcomm

Why consider experimental design?• If you’re performing experiments

• Cost • Time

• for experiment • for analysis

• Ethics • If you’re deciding to fund? to buy? to approve? to compete?

• are the results real? • can you trust the data?

Page 43: 2013 10-16-sbc3610-research methcomm

Main potential problems

• Pseudoreplication

• Confounding factors

• Insufficient data/power

• Inappropriate statistics

Wrong Inaccurate & Misleading