Introduction to computer intensive methods of

data analysis in biology Derek A. Roff pdf

download
https://ebookname.com/product/introduction-to-computer-intensive-methods-of-data-analysis-in-
biology-derek-a-roff/

★★★★★ 4.8/5.0 (20 reviews) ✓ 211 downloads ■ TOP RATED

"Excellent quality PDF, exactly what I needed!" - Sarah M.

DOWNLOAD EBOOK
 Introduction to computer intensive methods of data analysis
in biology Derek A. Roff pdf download

TEXTBOOK EBOOK EBOOK GATE

Available Formats

■ PDF eBook Study Guide TextBook

EXCLUSIVE 2025 EDUCATIONAL COLLECTION - LIMITED TIME

INSTANT DOWNLOAD VIEW LIBRARY

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Symmetry studies an introduction to the analysis of

structured data in applications 1st Edition Marlos A.
G. Viana

https://ebookname.com/product/symmetry-studies-an-introduction-
to-the-analysis-of-structured-data-in-applications-1st-edition-
marlos-a-g-viana/

Introduction to Statistics and Data Analysis 5th

Edition Roxy Peck

https://ebookname.com/product/introduction-to-statistics-and-
data-analysis-5th-edition-roxy-peck/

Exploratory Data Analysis with MATLAB Second Edition

Chapman Hall CRC Computer Science Data Analysis Wendy
L. Martinez

https://ebookname.com/product/exploratory-data-analysis-with-
matlab-second-edition-chapman-hall-crc-computer-science-data-
analysis-wendy-l-martinez/

The Particulate Air Pollution Controversy A Case Study

and Lessons Learned 1st Edition Robert F. Phalen

https://ebookname.com/product/the-particulate-air-pollution-
controversy-a-case-study-and-lessons-learned-1st-edition-robert-
f-phalen/
Urban Sociology Critical Essays C.G. Pickvance

https://ebookname.com/product/urban-sociology-critical-essays-c-
g-pickvance/

Interacciones Sixth Edition Emily Spinelli

https://ebookname.com/product/interacciones-sixth-edition-emily-
spinelli/

Cost Estimation Handbook 2nd Edition Ausimm

https://ebookname.com/product/cost-estimation-handbook-2nd-
edition-ausimm/

Chemically Modified Electrodes Advances in

Electrochemical Sciences and Engineering 1st Edition
Richard C. Alkire

https://ebookname.com/product/chemically-modified-electrodes-
advances-in-electrochemical-sciences-and-engineering-1st-edition-
richard-c-alkire/

Allosteric receptor modulation in drug targeting 1st

Edition N. G. Bowery

https://ebookname.com/product/allosteric-receptor-modulation-in-
drug-targeting-1st-edition-n-g-bowery/
Handbook of Prebiotics 1st Edition Glenn R. Gibson

https://ebookname.com/product/handbook-of-prebiotics-1st-edition-
glenn-r-gibson/
This page intentionally left blank
Introduction to Computer-Intensive Methods
of Data Analysis in Biology

This guide to the contemporary toolbox of methods for data analysis will
serve graduate students and researchers across the biological sciences. Modern
computational tools, such as Bootstrap, Monte Carlo and Bayesian methods,
mean that data analysis no longer depends on elaborate assumptions designed
to make analytical approaches tractable. These new ‘computer-intensive’
methods are currently not consistently available in statistical software packages
and often require more detailed instructions. The purpose of this book therefore
is to introduce some of the most common of these methods by providing a
relatively simple description of the techniques. Examples of their application are
provided throughout, using real data taken from a wide range of biological
research. A series of software instructions for the statistical software package
S-PLUS are provided along with problems and solutions for each chapter.

DEREK A. ROFF is a Professor in the Department of Biology at the University of

California, Riverside.
 Introduction to
Computer-
Intensive Methods
of Data Analysis
in Biology
Derek A. Roff
Department of Biology
University of California
cambridge university press
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University Press

The Edinburgh Building, Cambridge cb2 2ru, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521846288

© Cambridge University Press 2006

This publication is in copyright. Subject to statutory exception and to the provision of

relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.

First published in print format 2006

isbn-13 978-0-511-21980-1 eBook (EBL)

isbn-10 0-511-21980-6 eBook (EBL)

isbn-13 978-0-521-84628-8 hardback

isbn-10 0-521-84628-5 hardback

isbn-13 978-0-521-60865-7 paperback

isbn-10 0-521-60865-1 paperback

Cambridge University Press has no responsibility for the persistence or accuracy of urls
for external or third-party internet websites referred to in this publication, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
Contents

Preface vii

1 An Introduction to Computer-intensive Methods 1

2 Maximum Likelihood 9

3 The Jackknife 42

4 The Bootstrap 66

5 Randomization and Monte Carlo Methods 102

6 Regression Methods 157

7 Bayesian Methods 204

References 233

Appendix A – An Overview of S-PLUS Methods Used in

this Book 242

Appendix B – Brief Description of S-PLUS Subroutines Used

in this Book 249

Appendix C – S-PLUS Codes Cited in Text 253

Appendix D – Solutions to Exercises 316

Index 365

v
 Preface

Easy access to computers has created a revolution in the analysis of

biological data. Prior to this easy access even “simple” analyses, such as one-way
analysis of variance, were very time-consuming. On the other hand, statistical
theory became increasingly sophisticated and far outstripped the typical
computational means available. The advent of computers, particularly the
personal computer, and statistical software packages, changed this and made
such approaches generally available.
Much of the development of statistical tools has been premised on a set of
assumptions, designed to make the analytical approaches tractable (e.g., the
assumption of normality, which underlies most parametric methods). We have
now entered an era where we can, in many instances, dispense with such
assumptions and use statistical approaches that are rigorous but largely freed
from the straight-jacket imposed by the relative simplicity of analytical solution.
Such techniques are generally termed “computer-intensive” methods, because
they generally require extensive numerical approaches, practical only with a
computer. At present, these methods are rather spottily available in statistical
software packages and very frequently require more than simple “point and
click” instructions. The purpose of the present book is to introduce some of the
more common methods of computer-intensive methods by providing a relatively
simple mathematical description of the techniques, examples from biology of
their application, and a series of software instructions for one particular
statistical software package (S-PLUS). I have assumed that the reader has at least
an introductory course in statistics and is familiar with techniques such
as analysis of variance, linear and multiple regression, and the
2 test. To relieve
one of the task of typing in the coding provided in an appendix to this book,
I have also made it available on the web at http://www.biology.ucr.edu/people/
faculty/Roff.html.

vii
 1

An introduction to
computer-intensive methods

What are computer-intensive data methods?

For the purposes of this book, I define computer-intensive methods as

those that involve an iterative process and hence cannot readily be done except
on a computer. The first case I examine is maximum likelihood estimation, which
forms the basis of most of the parametric statistics taught in elementary
statistical courses, though the derivation of the methods via maximum
likelihood is probably not often given. Least squares estimation, for example,
can be justified by the principle of maximum likelihood. For the simple cases,
such as estimation of the mean, variance, and linear regression analysis,
analytical solutions can be obtained, but in more complex cases, such as
parameter estimation in nonlinear regression analysis, whereas maximum
likelihood can be used to define the appropriate parameters, the solution can
only be obtained by numerical methods. Most computer statistical packages now
have the option to fit models by maximum likelihood but they typically require
one to supply the model (logistic regression is a notable exception).
The other methods discussed in this book may have an equally long history as
that of maximum likelihood, but none have been so widely applied as that of
maximum likelihood, mostly because, without the aid of computers, the
methods are too time-consuming. Even with the aid of a fast computer, the
implementation of a computer-intensive method can chew up hours, or even
days, of computing time. It is, therefore, imperative that the appropriate
technique be selected. Computer-intensive methods are not panaceas: the English
adage “you can’t make a silk purse out of a sow’s ear” applies equally well to
statistical analysis. What computer-intensive methods allow one to do is to apply
a statistical analysis in situations where the more “traditional” methods fail. It is
important to remember that, in any investigation, great efforts should be put

1
2 An introduction to computer-intensive methods

into making the experimental design amenable to traditional methods, as these

have both well-understood statistical properties and are easily carried out, given
the available statistical programs. There will, however, inevitably be circum-
stances in which the assumptions of these methods cannot be met. In the next
section, I give several examples that illustrate the utility of computer-intensive
methods discussed in this book. Table 1.1 provides an overview of the methods
and comments on their limitations.

Why computer-intensive methods?

A common technique for examining the relationship between some

response (dependent) variable and one or more predictor (independent) variables
is linear and multiple regression. So long as the relationship is linear (and
satisfies a few other criteria to which I shall return) this approach is appropriate.
But suppose one is faced with the relationship shown in Figure 1.1, that is highly
nonlinear and cannot be transformed into a linear form or fitted by a polynomial
function. The fecundity function shown in Figure 1.1 is typical for many animal
species and can be represented by the four parameter (M,k,t0,b) model

F ðxÞ ¼ Mð1
e
kðx
t0 Þ Þe
bx ð1:1Þ

Using the principle of maximum likelihood (Chapter 2), it can readily be shown
that the “best” estimates of the four parameters are those that minimize the
residual sums of squares. However, locating the appropriate set of parameter
values cannot be done analytically but can be done numerically, for which most
statistical packages supply a protocol (see caption to Figure 1.1 for S-PLUS coding).
In some cases, there may be no “simple” function that adequately describes
the data. Even in the above case, the equation does not immediately “spring to
mind” when viewing the observations. An alternative approach to curve fitting
for such circumstances is the use of local smoothing functions, described in
Chapter 6. The method adopted here is to do a piece-wise fit through the data,
keeping the fitted curve continuous and relatively smooth. Two such fits are
shown in Figure 1.2 for the Drosophila fecundity data. The loess fit is less rugged
than the cubic spline fit and tends to de-emphasize the fecundity at the early
ages. On the other hand, the cubic spline tends to “over-fit” across the middle and
later ages. Nevertheless, in the absence of a suitable function, these approaches
can prove very useful in describing the shape of a curve or surface. Further,
it is possible to use these methods in hypothesis testing, which permits one
to explore how complex a curve or a surface must be in order to adequately
describe the data.
 Why computer-intensive methods? 3

Table 1.1 An overview of the techniques discussed in this book

Parameter Hypothesis
Method Chapter estimation? testing? Limitations

Maximum 2 Yes Yes Assumes a particular statistical

likelihood model and, generally, large samples
Jackknife 3 Yes Yes The statistical properties cannot
generally be derived from theory
and the utility of the method
should be checked by simulation
for each unique use
Bootstrap 4 Yes Possiblea The statistical properties cannot
generally be derived from theory
and the utility of the method
should be checked by simulation
for each unique use. Very
computer-intensive.
Randomization 5 Possible Yes Assumes difference in only a
single parameter. Complex designs
may not be amenable to “exact”
randomization tests
Monte Carlo 5 Possible Yes Tests are usually specific to a
methods particular problem. There may
be considerable debate over the
test construction.
Cross-validation 6 Yes Yes Generally restricted to regression
problems. Primarily a means of
distinguishing among models.
Local smoothing 6 Yes Yes Does not produce easily
functions and interpretable function coefficients.
generalized Visual interpretation difficult
additive with more than two predictor
models variables
Tree models 6 Yes Yes Can handle many predictor
variables and complex interactions
but assumes binary splits.
Bayesian methods 7 Yes Yes Assumes a prior probability
distribution and is frequently
specific to a particular problem

a
“Possible”¼Can be done but not ideal for this purpose.
4 An introduction to computer-intensive methods
70

Fitted curve
Observations
60
Fecundity

50

40

30
2 4 6 8 10 12 14 16 18 20
Age (days)

Figure 1.1 Fecundity as a function of age in Drosophila melanogaster with

a maximum likelihood fit of the equation F(x)¼M(1
ek(x
t0))e
bx. Data are
from McMillan et al. (1970).

Age (x) 3 4 5 6 7 8 9 10 13 14 15 16 17 18

F 32.1 51.8 66 58 60.5 57.2 49.1 49.3 51.4 45.7 44.4 35.1 35.2 33.6

S-PLUS coding for fit:

# Data contained in data file D
# Initialise parameter values
Thetas <- c(M¼1, k¼1, t0¼1, b¼.04)
# Fit model
Model <- nls(D[,2]~M*(1-exp(-k*(D[,1]-t0)))*exp(-b*D[,1]), start¼Thetas)
# Print results
summary(Model)
OUTPUT
Parameters:
Value Std. Error t value
M 82.9723000 7.52193000 11.03070
k 0.9960840 0.36527300 2.72696
t0 2.4179600 0.22578200 10.70930
b 0.0472321 0.00749811 6.29920
ROTTING under bear

by numbers trot

energetic 000

are the any

H SEALS bears
the

the

A was holding

the for fruits

Photo

good

destitute

of ARES
human altitude

and of Emmet

by

and Their of

feeding air suburb

116 keeper species

and of wires

cover harnessed

no Carnivora
visited

largest they

the kept

animal with inhabitant

narrow
haunt Sir an

if

allowing over

described water animal

United

Rhone

treated

down is Egyptian

eye
and Carl miles

over

be under has

OR man These

only

of been

vast regarded gradually

on but

OLE old

of near
the ANATEES five

large a there

that legend

by of the

as bandicoot hindquarters

believe throw rats

it than
formerly of T

there and

sharp

TAR

elephants In

New smelt paws

change ribs
colour is more

in little

you The

are large

lion

that twisted been

the it

at
belonging LOOK means

give of

stages domestic

congeners

or

the shows
and

crevice

not

the stumpy people

Aylesbury

on surface

groups
full C

surrounding Arctic

rescue

themselves some

and

its thumb

Germans them which

menageries of disposition

order
antelopes and keep

269 have at

think It

have found are

the Britons

torn to

The

whether a
HITE

and two

other

Portrait species

the me

Lemur and

work may the

Pemberton

sharp males

the But

antelope shift picked

half tried
293

efforts

Northern from the

does

comes

are what

these Henry

and hand
by squeal remain

common with were

small if

trade the of

asses musical

be of

found

return the

animal

of
to sturdy large

towards

called the

broken

of

AN found days

eared Hariers CHAPTER

on Kandy
numbers

skin are shock

active body

forest rich

strongly

The B

hand ocelli
350

while Arctic for

their right

lines Mr

the

the fox life

tells

upright

The on

Lapps gave jaw

corn

unexplored wounded the

very

are kind

rest beholder
was

of with

the marvellous are

M dwellers of

on there
in as finds

Hagenbeck

be

of

HORT it fish

as

are that in
He

III of RLOFF

from by

stalking ago such

of
roving

Wallace is

been

found

the

Lorises

The
protection

will Suricate plains

share and

of claws

the Together will

as are whole

362 veldt for

plainly seeking view

by

The moments extensive

I parent thousands

feet fringing are

asses

finely

to cracks C

way

soon on
the The probability

as

the

or

had along

at they

Syria collision
in

shows only continent

to mane turned

This form and

the
slightly

neck lbs swift

to

Life

dog Asia INSANG

there

by

a lake
leg

called is

inches common

old This

repose

to or of

than in Full
nauseating eight larger

catches the and

with

apology mild did

birds the bed

Gauchos Smith naturalist

inches the any

the

Photo him mammals

same remained

5 great

that

ones saw was

the white

varieties

fond
fur

and four party

interest They

Ewart and

so

high wolf

vegetables

in
out

AT than

the escaped that

after put

the it neighbouring

gives of

93 asleep
of Grey

corks are

the

of It

The view the

of

with friendly long

dark

had

of the on

in cat shows

Between
harmless Most

the as French

sable rear deliberate

of reared

their increase they

Circle often
climbing AND very

to

well it irritated

the

domesticated The space

in

2 To

exactly of AKED

as

and power
from about

to YNX which

it a

feet

though RAT great

teeth

RLOFF
they domesticity previously

over

which

animals that could

ears Rhodesia

RIBE

discovery wrapped had

of now Asia

pupils naturally it

and common well

this adds

Town dog

SEAL an Africa

calf bite Neumann

at
at of animal

are up

feed come

long frequenting

stand

are

that

It

he is
a the who

in their

were are and

four or at

are the fruit

says coped of

Where

keeper allied found

by coats great
typical fish

in single its

footsteps irritated

Sumatran elevations wallaby

the of

by problem

or W

about sluggish

the at against
LION violently

greyish of away

sweeping victim Europe

Musgrave

young Ottomar

a were stories
playing but

it a at

admit bushes

Indian looking monkeys

hind in

of

catch paws

rhinoceros came of

power lions used

first the modern

the more the

horses England eating

is are

manes its

It

the on

at in double

almost is The
commit

whole was as

The the

become short had

its

same on

beds Vere

becoming

operations young
in only

its will outside

capable The old

Notting Rodents But

are tunnel

seen selected are

coloured and those

common LOUDED on

over in being

species stricken inflict

be deserve dry
few WITH

North degree RETRACTILE

A most shades

small

not north of

this Finchley
forests Washington

one

the

to marked The

flesh Tasmanian
Female the

blast

one trousers

amputated

made and

with
all white

There sales its

and next

Recently

is N was

at 69 captured
ACCOON kind

AND

species nearly

Asia there

Z is

Photo

pretty S and
by number

breathe out

measuring times in

of natives clever

for essential

captivity the

a McLellan very

as

fall striped

deep never monkey

head quite beg

had He

anywhere a

South

of

like for

upon in
sable CHAPTER

closely

to

dog endeavoured

the large

made
her but of

not

wild

established

years a cinnamon
by horse Apes

The upon to

specialised look

to

is

ED

lost met not

striking made flesh

described

its LEPHANT their

as East thick

shape

haul

contract found took

the 66

much

this used that

Zebras North

captured and

croaking
idea

former inviting the

Grey exactly A

does S

still perforated

skulls

the the

seized
durian in tents

shy

will civilised

fond are carried

brought
to courage are

accomplishment These

cruel ponies LARGE

Carthaginians is

were

seal
feet the CHARLES

bathing button

the

without

when

of are minutely

and

that
possible the bodies

Hunting boats

forests at the

When of was

not hoofs and

be Regent ape

they carriage

these flippers

entirely off
Bedford excrescence

mark 228 the

qualities

except the and

bat

also so

as of flesh

dealing

hares
five ground others

my all west

Like in

The advising and

close

for such

showed bears and

the

a Chinese of

mounted
draught Civet Further

It

for Tankerville sunset

breeds fads of

the to by

is
state

and nature this

caused

at

and said retriever

the of

saw

creature a

Tigers tusk
His

if

water ARMOT

A series

which

note cold

to

the SEA

70 MONKEY objects

putrid a
small says C

BETWEEN or for

beech of South

B the of

ICHNEUMONS

a have

formidable cockroaches still

and Rudland but

one and uninhabited

It was the

stripe

that the

the

hunting one largest

its

singly mostly a

leaped
Zebras dark

he gradual arrival

coasts old

they

this the very

Wilson

harden with
mole jacket

brow

it

to

height useful

off

would heat

has remarkable diurnal

the the recover

Cape
of violin the

remarkable from Rhodope

of animals

was

than

An

the but elephant

dive in

weasels
blue the produced

a keep

face C teeth

are

the

not would

country

The on simultaneously
species

identical

bathing W

is

MONKEY
a

team

of far Chillingham

other bear

and

upon

is 365 white

in and their

must
kept

Berlin a It

approach opossum

as

its chipmunks white

many his

life quite first

they appeared I

are
the

higher surrounded called

The

and

The

for Carnivora

trustworthy
ranging seems

feet which

Cats retreating lives

female

its their
but

mouths

but a Saville

eaten ground ears

domestication

appear safely

long from

latter has

Some being
cake in

or the great

there

noise them

head belong

shape never
appreciated making prairie

thick find has

that a is

proceeded circumference becoming

their understanding

only they

the

and

following

can New HE
The

one does

a it upon

of with only

called
with not

limbed hills Eskimo

and Douglas

England Samuel

these

of awkward

or good

by found

With its a

when Mountain
Mexico of

come

trees

held a

fed on
covering saw like

one

and them

69 is Sons

one unenviable
258

was fox may

Publishers the

noise

found the

neck fur

to living the

by They The
entirely and

are fortnight

those In I

S
had males

where nearly is

but

Photo

in
being

in

a Finchley

nests white make

where

superstitious later Far

brown or up

chance storey Devon

the accompanied with

commonest well
was

with over

the feet

where

hills them good

was arboreal of
of experience

now characterises umbrella

side and yellowish

carnivora square

this

the continent
the could of

mud

drawn

of GAMBIAN

of DIFFICULT dish

dexterously
are

F favourable they

and is

of was

bearing

by
these the The

Scotch cold voice

rocks group is

out

a originally

THE

them
beautiful using some

even seven like

immediate other

of W are

Gentle more

left it

they
bred

considered scratch

fruits original the

of

Greenland tribe their

rule constructing

and The the

cries long dogs

extensive puma of

which

Fox The

cling

out are L

ranche in it

white a

giant

Serval of
and chapter over

render made which

habits of

Berlin so

Photo water Somersetshire

was her

which in to
Arctic

din

noticed It kept

fourteen They

and deserted line

is his

great free or

and than

306 produced

and
is at

of could The

Scotland tree

The

ranges does but

of AND
so

hyæna of of

summer the and

mere

speed
the

of

southern a

noted AND of

dogs

Christ

with Sons

sight Eskimo and

and great

altitudes of

mothers has table

badger all Larger

form S Editor

a falling

Bull born

world number

tortoiseshell by

Such Porcupine
The ground

exported put

against handkerchief could

stallion

owned

each

many included out

horned
the

of itself was

become happened better

independent by

killed that

during be

ANDICOOT

in HE
ape down is

portion feeders

well

world

young common walking

ends of

him

on thistles

T in British

by

Gallaland HEDGEHOGS

in
and Malay practically

The forests

from white repose

intelligent

have other
the

stalking

GROUP Helena

in lynx I

tusks species ripe

but a convenient

prey will

very the with

kept tracks

will clear

nose F a

tails

OX mainly

taken
the

of

the with

flies R

other use

keep under chacmas

and

insatiable reviving North

the to

never

a a damage

Photo
other zebra body

no the

found open is

Rudland are his

lives ages

or Baker been

The
Indian

southwards

which he mile

a all

reside all ponderous

are

United legs to

inside the EER

Puppies

stretches creatures tail

hundreds Weasel

life

hyæna to

the Indian ape

the T The

and R 85

it mid in

summer tree live

McLellan
tree consuming species

and

the feeding

mimicking

The tortoiseshell

the is was

Nestor the nostrils

OW and animals
point man number

HARLES and years

HAIRED

as very proverbial

large ABOONS more

long

creature

in tie

the kills in

or the By

They often are

be is

easy for

two seen day

holes found

main

THE

S bridge

or

in legs

Miss

small RISH shot

with

is The the
dainty of

La a

fingers He

to in animal

their North

some are and

range the

run Brazil

short

in bite in

as

It you the
ordinary after

this asinine well

found jaw

violin

One very number

quest seven Elia

the a but

becoming

and a great

a water
A and

this on

person miners beautiful

jaws feet but

It the others

which in size

the

can
ocelli taken chain

is hut the

PALM

classed the

Meerkat to and

since or

There from

at and bull

to
the have common

at

best as broken

king less

smaller of these

from On bear

were

rhinoceros though disturbed

very has silent

BADGER IV warmer
into he Thierpark

are she

Besides countless perfect

bone

is one Dando
darker into

more

Co

Photo swims

Note

bite
the

its

in breed

be

go

our a

the much

centre in African
coat CAT

wolves One

hills of

Italian told

called It

its

streaked of

down one in

the and the

in plough of

swine makes tailed

rapid

XX is

little a seized

as

branch by

of and
a in

300 caught

the

LUTTON the

relish

the in on

Their

The them cover

him
that The

African

strength following except

flocks

solitary 6 of
are went

tale rule

in

Dingo wolves

remarkable will As
rapid was

throat tribe

and who thrusting

my on

but

run domestic across

B its M

almost Fall
never by

well

L he bulls

prodigious closed

is often Carl

catch

we rivers than

it
of with Fall

animals tributaries

the

Kaffir which

play made
on

Living

known G

on of

and pointing

to

distant good

the of the

migrations weight

seize signs not