Computational Cell Biology Methods and Protocols

Louise Von Stechow download

https://textbookfull.com/product/computational-cell-biology-
methods-and-protocols-louise-von-stechow/

Download more ebook from https://textbookfull.com

We believe these products will be a great fit for you. Click
the link to download now, or visit textbookfull.com
to discover even more!

Cancer Systems Biology Methods and Protocols 1st

Edition Louise Von Stechow (Eds.)

https://textbookfull.com/product/cancer-systems-biology-methods-
and-protocols-1st-edition-louise-von-stechow-eds/

Computational Stem Cell Biology Methods and Protocols

Patrick Cahan

https://textbookfull.com/product/computational-stem-cell-biology-
methods-and-protocols-patrick-cahan/

Stem Cell Renewal and Cell Cell Communication Methods

and Protocols Methods in Molecular Biology 2346 Kursad
Turksen Editor

https://textbookfull.com/product/stem-cell-renewal-and-cell-cell-
communication-methods-and-protocols-methods-in-molecular-
biology-2346-kursad-turksen-editor/

Epithelial Cell Culture Methods and Protocols Methods

in Molecular Biology 2749 Mario Baratta

https://textbookfull.com/product/epithelial-cell-culture-methods-
and-protocols-methods-in-molecular-biology-2749-mario-baratta/
3D Cell Culture Methods and Protocols Methods in
Molecular Biology 2764 2nd Edition Zuzana Sumbalova
Koledova

https://textbookfull.com/product/3d-cell-culture-methods-and-
protocols-methods-in-molecular-biology-2764-2nd-edition-zuzana-
sumbalova-koledova/

Single Cell Analysis: Methods and Protocols Miodrag

Gužvi■

https://textbookfull.com/product/single-cell-analysis-methods-
and-protocols-miodrag-guzvic/

T Cell Development Methods and Protocols 1st Edition

Rémy Bosselut

https://textbookfull.com/product/t-cell-development-methods-and-
protocols-1st-edition-remy-bosselut/

Systems Chemical Biology: Methods and Protocols Slava

Ziegler

https://textbookfull.com/product/systems-chemical-biology-
methods-and-protocols-slava-ziegler/

Cell-free DNA as Diagnostic Markers: Methods and

Protocols Valentina Casadio

https://textbookfull.com/product/cell-free-dna-as-diagnostic-
markers-methods-and-protocols-valentina-casadio/
Methods in
Molecular Biology 1819

Louise von Stechow

Alberto Santos Delgado
Editors

Computational
Cell Biology
Methods and Protocols
M E THODS IN M OLECULAR B IOLOGY

Series Editor
John M. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, AL10 9AB, UK

For further volumes:

http://www.springer.com/series/7651
 Computational Cell Biology

Methods and Protocols

Edited by

Louise von Stechow

NNF Center for Protein Research, University of Copenhagen, Copenhagen, Denmark

Alberto Santos Delgado

Novo Nordisk Foundation Center for Protein Research, University of Copenhagen, Copenhagen, Denmark
Editors
Louise von Stechow Alberto Santos Delgado
NNF Center for Protein Research Novo Nordisk Foundation Center for Protein
University of Copenhagen Research
Copenhagen, Denmark University of Copenhagen
Copenhagen, Denmark

ISSN 1064-3745 ISSN 1940-6029 (electronic)

Methods in Molecular Biology
ISBN 978-1-4939-8617-0 ISBN 978-1-4939-8618-7 (eBook)
https://doi.org/10.1007/978-1-4939-8618-7

Library of Congress Control Number: 2018956725

© Springer Science+Business Media, LLC, part of Springer Nature 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed
to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been
made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Cover of the book was designed by Dr. Francesco Russo – Faculty of Health and Medical Sciences, Novo Nordisk
Foundation Center for Protein Research, University of Copenhagen, Copenhagen, Denmark

This Humana Press imprint is published by the registered company Springer Science+Business Media, LLC part of
Springer Nature.
The registered company address is: 233 Spring Street, New York, NY 10013, U.S.A.
Preface

Technological advances over the past decade have resulted in an explosion of available data,
putting an end to researchers’ focus on single genes or proteins and promoting system-wide
approaches into biomedical research. The so-called big data era brings along the need for
ways to extract meaningful information that go beyond manual inspection of large-scale
datasets. An expanding toolbox of computational methods is evolving for identification
and interpretation of biological phenotypes. Data-driven analyses, gene and protein set
enrichment, representation of large-scale data into networks, and mathematical modeling
of biological phenotypes are now emerging as means for the sophisticated analysis of the
available biological data.
Computational Cell Biology: Methods and Protocols is targeted toward scientists who wish
to employ computational techniques for analyses of a wide range of biological contexts,
providing a great overview of suitable methods currently used in the field. It is written for a
broad audience ranging from researchers who are unfamiliar with computational biology to
those with more experience in the field. A number of review-style chapters give an overview
of available data resources and analysis methods, while easy-to-follow protocols allow the
researcher to apply various computational tools to an array of different data types.

Copenhagen, Denmark Louise von Stechow

Alberto Santos Delgado

v
Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

PART I BIG DATA- AND I TS I MPLICATIONS IN CELL BIOLOGY

1 Rule-Based Models and Applications in Biology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Álvaro Bustos, Ignacio Fuenzalida, Rodrigo Santibáñez,
Tomás Pérez-Acle, and Alberto J.M. Martin
2 Optimized Protein–Protein Interaction Network Usage with Context
Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Natalia Pietrosemoli and Maria Pamela Dobay

PART II DATA-DRIVEN ANALYSES OF HIGH-THROUGHPUT DATASETS

3 SignaLink: Multilayered Regulatory Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

Luca Csabai, Márton Ölbei, Aidan Budd, Tamás Korcsmáros,
and Dávid Fazekas
4 Interplay Between Long Noncoding RNAs and MicroRNAs in Cancer . . . . . . . . . 75
Francesco Russo, Giulia Fiscon, Federica Conte, Milena Rizzo, Paola Paci,
and Marco Pellegrini
5 Methods and Tools in Genome-wide Association Studies . . . . . . . . . . . . . . . . . . . . . . . . 93
Anja C. Gumpinger, Damian Roqueiro, Dominik G. Grimm,
and Karsten M. Borgwardt

PART III NETWORK-BASED MODELING OF CELLULAR PHENOTYPES

6 Identifying Differentially Expressed Genes Using Fluorescence-Activated

Cell Sorting (FACS) and RNA Sequencing from Low Input Samples . . . . . . . . . . . 139
Natalie M. Clark, Adam P. Fisher, and Rosangela Sozzani
7 Computational and Experimental Approaches to Predict Host–Parasite
Protein–Protein Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Yesid Cuesta-Astroz and Guilherme Oliveira
8 An Integrative Approach to Virus–Host Protein–Protein Interactions . . . . . . . . . . . 175
Helen V. Cook and Lars Juhl Jensen
9 The SQUAD Method for the Qualitative Modeling of Regulatory Networks . . . 197
Akram Méndez, Carlos Ramírez, Mauricio Pérez Martínez,
and Luis Mendoza
10 miRNet—Functional Analysis and Visual Exploration of miRNA–Target
Interactions in a Network Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Yannan Fan and Jianguo Xia

vii
viii Contents

11 Systems Biology Analysis to Understand Regulatory miRNA Networks

in Lung Cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Meik Kunz, Andreas Pittroff, and Thomas Dandekar
12 Spatial Analysis of Functional Enrichment (SAFE) in Large Biological
Networks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Anastasia Baryshnikova

PART IV MATHEMATICAL MODELING OF CELLULAR PHENOTYPES

13 Toward Large-Scale Computational Prediction of Protein Complexes . . . . . . . . . . . 271

Simone Rizzetto and Attila Csikász-Nagy
14 Computational Models of Cell Cycle Transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
Rosa Hernansaiz-Ballesteros, Kirsten Jenkins, and Attila Csikász-Nagy
15 Simultaneous Profiling of DNA Accessibility and Gene Expression
Dynamics with ATAC-Seq and RNA-Seq. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
David G. Hendrickson, Ilya Soifer, Bernd J. Wranik, David Botstein,
and R. Scott McIsaac
16 Computational Network Analysis for Drug Toxicity Prediction . . . . . . . . . . . . . . . . . . 335
C. Hardt, C. Bauer, J. Schuchhardt, and R. Herwig
17 Modeling the Epigenetic Landscape in Plant Development . . . . . . . . . . . . . . . . . . . . . . 357
Jose Davila-Velderrain, Jose Luis Caldu-Primo,
Juan Carlos Martinez-Garcia, and Elena R. Alvarez-Buylla
18 Developing Network Models of Multiscale Host Responses Involved
in Infections and Diseases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
Rohith Palli and Juilee Thakar

PART V COMPUTATIONAL ANALYSES OF HETEROGENOUS CELL

POPULATIONS

19 Exploring Dynamics and Noise in Gonadotropin-Releasing Hormone

(GnRH) Signaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
Margaritis Voliotis, Kathryn L. Garner, Hussah Alobaid,
Krasimira Tsaneva-Atanasova, and Craig A. McArdle

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
Contributors

HUSSAH ALOBAID • Laboratories for Integrative Neuroscience and Endocrinology, School of

Clinical Sciences, University of Bristol, Bristol, UK
ELENA R. ALVAREZ-BUYLLA • Laboratorio de Genética Molecular, Desarrollo y Evolución
de Plantas, México, México; Instituto de Ecología, Centro de Ciencias de la Complejidad
(C3), Universidad Nacional Autónoma de México Ciudad Universitaria, México, México
ANASTASIA BARYSHNIKOVA • Lewis-Sigler Institute for Integrative Genomics, Princeton
University, Princeton, NJ, USA; Calico Life Sciences LLC, South San Francisco, CA, USA
C. BAUER • MicroDiscovery GmbH, Berlin, Germany
KARSTEN M. BORGWARDT • Machine Learning and Computational Biology Lab, D-
BSSE, ETH Zurich, Basel, Switzerland; SIB Swiss Institute of Bioinformatics, Lausanne,
Switzerland
DAVID BOTSTEIN • Calico Life Sciences, South San Francisco, CA, USA
AIDAN BUDD • Earlham Institute, Norwich Research Park, Norwich, UK
ÁLVARO BUSTOS DELGADO • Computational Biology Lab, Fundacion Ciencia & Vida,
Santiago, Chile
JOSE LUIS CALDU-PRIMO • Centro de Ciencias de la Complejidad (C3), Universidad
Nacional Autónoma de México, Ciudad Universitaria, México, México
NATALIE M. CLARK • Department of Plant and Microbial Biology, North Carolina State
University, Raleigh, NC, USA; Biomathematics Graduate Program, North Carolina State
University, Raleigh, NC, USA
FEDERICA CONTE • Institute for Systems Analysis and Computer Science “A. Ruberti”
(IASI), National Research Council (CNR), Rome, Italy
HELEN V. COOK • Novo Nordisk Center for Protein Research, University of Copenhagen,
Copenhagen, Denmark
LUCA CSABAI • Eötvös Loránd University, Budapest, Hungary
ATTILA CSIKÁSZ-NAGY • Randall Centre for Cell and Molecular Biophysics and Institute
for Mathematical and Molecular Biomedicine, King’s College London, London, UK;
Faculty of Information Technology and Bionics, Pázmány Péter Catholic University,
Budapest, Hungary
YESID CUESTA-ASTROZ • Centro de Pesquisas René Rachou (CPqRR), Fundação Oswaldo
Cruz (FIOCRUZ), Belo Horizonte, Minas Gerais, Brazil
THOMAS DANDEKAR • Department of Bioinformatics, Functional Genomics and Systems
Biology Group, Biocenter, Würzburg, Germany; BioComputing Unit, EMBL Heidelberg,
Heidelberg, Germany
JOSE DAVILA-VELDERRAIN • Centro de Ciencias de la Complejidad (C3), Universidad
Nacional Autónoma de México, Ciudad Universitaria, México, México; Departamento
de Control Automático, Cinvestav-IPN, Cambridge, México D.F, Mexico; MIT Computer
Science and Artificial Intelligence Laboratory, Cambridge, MA, USA; •Broad Institute of
MIT and Harvard, Cambridge, MA, USA
MARIA PAMELA DOBAY • SIB Swiss Institute of Bioinformatics, Quartier Sorge, Bâtiment
Génopode, Lausanne, Switzerland; IQVIA, Basel, Switzerland; Yocto Group Limited,
Zurich, Switzerland

ix
x Contributors

YANNAN FAN • Institute of Parasitology, McGill University, Sainte Anne de Bellevue, QC,
Canada
DÁVID FAZEKAS • Eötvös Loránd University, Budapest, Hungary; Earlham Institute, Nor-
wich Research Park, Norwich, UK
GIULIA FISCON • Institute for Systems Analysis and Computer Science “A. Ruberti” (IASI),
National Research Council (CNR), Rome, Italy
ADAM P. FISHER • Department of Plant and Microbial Biology, North Carolina State
University, Raleigh, NC, USA
IGNACIO FUENZALIDA • Computational Biology Lab, Fundacion Ciencia & Vida, Santi-
ago, Chile
KATHRYN L. GARNER • Laboratories for Integrative Neuroscience and Endocrinology,
School of Clinical Sciences, University of Bristol, Bristol, UK
DOMINIK G. GRIMM • Machine Learning and Computational Biology Lab, D-BSSE, ETH
Zurich, Basel, Switzerland; SIB Swiss Institute of Bioinformatics, Lausanne, Switzerland
ANJA C. GUMPINGER • Machine Learning and Computational Biology Lab, ETH Zurich,
Basel, Switzerland; SIB Swiss Institute of Bioinformatics, Lausanne, Switzerland
C. HARDT • Department of Computational Molecular Biology, Max-Planck-Institute for
Molecular Genetics, Berlin, Germany
DAVID G. HENDRICKSON • Calico Life Sciences, South San Francisco, CA, USA
ROSA HERNANSAIZ-BALLESTEROS • Randall Division of Cell and Molecular Biophysics
and Institute for Mathematical and Molecular Biomedicine, King’s College London,
London, UK
R. HERWIG • Department of Computational Molecular Biology, Max-Planck-Institute for
Molecular Genetics, Berlin, Germany
KIRSTEN JENKINS • Randall Division of Cell and Molecular Biophysics and Institute for
Mathematical and Molecular Biomedicine, King’s College London, London, UK
LARS JUHL JENSEN • Novo Nordisk Center for Protein Research, University of Copen-
hagen, Copenhagen, Denmark
TAMÁS KORCSMÁROS • Eötvös Loránd University, Budapest, Hungary; Earlham Institute,
Norwich Research Park, Norwich, UK; Quadram Institute, Norwich Research Park,
Norwich, UK
MEIK KUNZ • Department of Bioinformatics, Functional Genomics and Systems Biology
Group, Biocenter, Würzburg, Germany
ALBERTO J. M. MARTIN • Computational Biology Lab, Fundacion Ciencia & Vida, San-
tiago, Chile; Centro Interdisciplinario de Neurociencias de Valparaiso, Valparaiso, Chile;
Centro de Genomica y Bioinformatica, Universidad Mayor, Santiago, Chile
MAURICIO PÉREZ MARTÍNEZ • Instituto de Investigaciones Biomédicas, Universidad
Nacional Autónoma de México, CDMX, Mexico, Mexico
JUAN CARLOS MARTINEZ-GARCIA • Departamento de Control Automático, Cinvestav-
IPN, México, México
CRAIG A. MCARDLE • Laboratories for Integrative Neuroscience and Endocrinology, School
of Clinical Sciences, University of Bristol, Bristol, UK
R. SCOTT MCISAAC • Calico Life Sciences, South San Francisco, CA, USA
AKRAM MÉNDEZ • Instituto de Investigaciones Biomédicas, Universidad Nacional
Autónoma de México, CDMX, Mexico, Mexico
LUIS MENDOZA • Instituto de Investigaciones Biomédicas, Universidad Nacional
Autónoma de México, CDMX, Mexico, Mexico
MÁRTON ÖLBEI • Earlham Institute, Norwich Research Park, Norwich, UK; Quadram
Institute, Norwich Research Park, Norwich, UK
 Contributors xi

GUILHERME OLIVEIRA • Instituto Tecnológico Vale, Belém, PA, Brazil

PAOLA PACI • Institute for Systems Analysis and Computer Science “A. Ruberti” (IASI),
National Research Council (CNR), Rome, Italy
ROHITH PALLI • Medical Scientist Training Program and Biophysics, Structural & Com-
putational Biology graduate program, Rochester, NY, USA
MARCO PELLEGRINI • Institute of Informatics and Telematics (IIT), National Research
Council (CNR), Pisa, Italy
TOMÁS PÉREZ-ACLE • Computational Biology Lab, Fundacion Ciencia & Vida, Santiago,
Chile; Centro Interdisciplinario de Neurociencias de Valparaiso, Valparaiso, Chile
NATALIA PIETROSEMOLI • Institut Pasteur, Bioinformatics and Biostatistics Hub, C3BI,
USR 3756 CNRS, Paris, France
ANDREAS PITTROFF • Department of Bioinformatics, Functional Genomics and Systems
Biology Group, Biocenter, Würzburg, Germany
CARLOS RAMÍREZ • Instituto de Investigaciones Biomédicas, Universidad Nacional
Autónoma de México, CDMX, Mexico, Mexico
SIMONE RIZZETTO • School of Medical Sciences, University of New South Wales, Sydney,
NSW, Australia; Viral Immunology Systems Program, Kirby Institute for Infection and
Immunity, University of New South Wales, Sydney, NSW, Australia
MILENA RIZZO • Institute of Clinical Physiology, National Research Council (CNR), Pisa,
Italy; Istituto Toscano Tumori (ITT), Firenze, Italy
DAMIAN ROQUEIRO • Machine Learning and Computational Biology Lab, D-BSSE, ETH
Zurich, Basel, Switzerland; SIB Swiss Institute of Bioinformatics, Lausanne, Switzerland
FRANCESCO RUSSO • Novo Nordisk Foundation Center for Protein Research, Faculty of
Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
RODRIGO SANTIBÁÑEZ • Computational Biology Lab, Fundacion Ciencia & Vida, San-
tiago, Chile; Escuela de Ingeniería, Pontificia Universidad Católica de Chile, Santiago,
Chile
J. SCHUCHHARDT • MicroDiscovery GmbH, Berlin, Germany
ILYA SOIFER • Calico Life Sciences, South San Francisco, CA, USA
ROSANGELA SOZZANI • Department of Plant and Microbial Biology, North Carolina State
University, Raleigh, NC, USA; Biomathematics Graduate Program, North Carolina State
University, Raleigh, NC, USA
JUILEE THAKAR • Department of Microbiology and Immunology, University of Rochester
Medical Center, Rochester, NY, USA; Department of Biostatistics and Computational
Biology, University of Rochester Medical Center, Rochester, NY, USA
KRASIMIRA TSANEVA-ATANASOVA • EPSRC Centre for Predictive Modeling in Health-
care, University of Exeter, Exeter, UK; Department of Mathematics and Living Systems
Institute, College of Engineering, Mathematics and Physical Sciences, University of Exeter,
Exeter, UK
MARGARITIS VOLIOTIS • EPSRC Centre for Predictive Modeling in Healthcare, Univer-
sity of Exeter, Exeter, UK; Department of Mathematics and Living Systems Institute,
College of Engineering, Mathematics and Physical Sciences, University of Exeter, Exeter,
UK
BERND WRANIK • Calico Life Sciences, South San Francisco, CA, USA
JIANGUO XIA • Institute of Parasitology, McGill University, Sainte Anne de Bellevue,
Quebec, Canada; Department of Animal Science, McGill University, Sainte Anne de
Bellevue, Quebec, Canada
 Part I

Big Data- and Its Implications in Cell Biology

 Chapter 1

Rule-Based Models and Applications in Biology

Álvaro Bustos, Ignacio Fuenzalida, Rodrigo Santibáñez,
Tomás Pérez-Acle, and Alberto J. M. Martin

Abstract
Complex systems are governed by dynamic processes whose underlying causal rules are difficult to unravel.
However, chemical reactions, molecular interactions, and many other complex systems can be usually
represented as concentrations or quantities that vary over time, which provides a framework to study these
dynamic relationships. An increasing number of tools use these quantifications to simulate dynamically
complex systems to better understand their underlying processes. The application of such methods covers
several research areas from biology and chemistry to ecology and even social sciences.
In the following chapter, we introduce the concept of rule-based simulations based on the Stochastic
Simulation Algorithm (SSA) as well as other mathematical methods such as Ordinary Differential Equa-
tions (ODE) models to describe agent-based systems. Besides, we describe the mathematical framework
behind Kappa (κ), a rule-based language for the modeling of complex systems, and some extensions for
spatial models implemented in PISKaS (Parallel Implementation of a Spatial Kappa Simulator). To facilitate
the understanding of these methods, we include examples of how these models can be used to describe
population dynamics in a simple predator–prey ecosystem or to simulate circadian rhythm changes.

Key words Stochastic simulation, Rule-based modeling, κ language

1 The Stochastic Simulation Algorithm (SSA)

The SSA, also known as Gillespie’s algorithm [10], is the basis

of most stochastic simulation tools available. This algorithm and
the tools based on it assume there is a homogeneous and “well-
stirred” system of particles named agents. Agents can represent
any type of entity within a system, i.e., molecules or individuals,
and the interactions between agents are determined by a set
of rules or equations taking place at certain rates. These rules
are ordered and divided into agents to which the rule applies
and products (outcome agents). For instance, in a system of
chemical reactions described by an equation or rule (reactants →
products), every set of particles matching the left side of the
equation (or reactant agents) has an equal probability of being

Louise von Stechow and Alberto Santos Delgado (eds.), Computational Cell Biology: Methods and Protocols,
Methods in Molecular Biology, vol. 1819, https://doi.org/10.1007/978-1-4939-8618-7_1,
© Springer Science+Business Media, LLC, part of Springer Nature 2018

3
4 Álvaro Bustos et al.

the subject of that rule, that is to undergo the process described

by the rule. To clarify, given a reaction of the form A + B → C
in a system with 1000 particles of type A and 1000 of type B,
this “well-stirring” assumption means that every pair of particles
{A1 , B1 }, {A1 , B2 }, . . . , {Ai , Bj }, . . . , {A1000 , B1000 } has equal prob-
ability of interacting to produce a particle of type C. Another
important assumption made by the SSA is that the volume or
area where the simulation takes place is fixed, and thus, concen-
trations of agents correspond to the discrete number of agents of
each type.
To describe chemical systems, and this can be extended to
any other type of system, a specific set of reactions is required.
Reactions in this algorithm always match the following schema
(Eq. (1)):

m1 A1 + · · · + mr Ar → n1 C1 + · · · + ns Cs (1)

Whenever a reaction of this type takes place, a set of m reactant

particles of types Ai are removed from the simulation (for i =
1, . . . , r) and are in turn replaced by another set of n product
particles of types Cj (for j = 1, . . . , s). It should be noted that
any of the products Cj could be of the same kind as one of
the reactants, and that this schema covers reactions as simple as
Ai → Aj to more complicated reactions requiring several types of
different agents. Other important reactions that follow the same
schema are Ai → Ø to indicate degradation of agents and very
similarly, Ø → Ai to model the addition of a new element in
a system. Rules are applied according to reaction rates, which
defines different behaviors of the system upon variations in the
concentration of its reactants.
The quantity of each type of agent, or state of the system, at a
given time t can be represented by a vector of non-negative inte-
gers, or state vector, in which each entry represents the amount of
each agent type. The outcome of a given chemical reaction can also
be represented by a state-change vector, with the same size as the
state-vector at time t. The negative entries in the state-vector depict
the consumption of an agent, positive values mean the creation of
an agent, and 0 or null value indicates no change for a particular
agent type. Therefore, if the state vector before a given reaction
r is X  and the associated state-change vector is dr , the state of
the system changes from X  to X + dr after the reaction occurs.
For example, in a medium that has samples of three different
chemicals A, B, C, the following vector represents the existence of
1000 molecules of type A, 900 of type B, and 1200 of type C at
time t:
 Rule-Based Models and Applications in Biology 5

⎡ ⎤
1000

X(t) = ⎣ 900 ⎦
1200

In a similar way, the following two chemical reactions r1 and r2 cor-

respond, respectively, to the two state-change vectors dr1 and dr2 :
r1 r2
2A + B −→C 2C −→ 2A + B + C
⎡ ⎤ ⎡ ⎤
−2 2
 ⎣
dr1 = −1⎦  ⎣
dr2 = 1 ⎦
1 −1

Note that in the second reaction one of the particles of type C

acts as a catalyst and the outcome effect of reaction r2 is the same
→ 2A + B with a dr2 state-change vector identical to
r2
as of C −
−dr1 . However, the relationship between the different probabilities
of reactions r1 and r2 happening and the amount of particles
of type C present lead to different long-term behaviors of the
system.
A reaction r is fully specified by the state-change vector dr
and a propensity function a. This propensity function takes the
state vector X as argument and calculates the rate of every reaction
r in the system; thus, if a(r1 , X) > a(r2 , X)
 reaction r1 is more
likely to occur than r2 . This is a discrete model; therefore, the
 is combinatorial in nature. For a fixed r, it should,
function a(r, X)
theoretically, be directly proportional to the number of distinct sets
of molecules that match the left side of the equation describing the
reaction r and the physical properties of the medium being sim-
ulated. In this way, a probabilistic mathematical model of any set
of reactions can be built given their state-change vectors dr and a
propensity function a. The function a reflects the constraints given
by the chemical nature of the system being modeled and allows the
description, at least indirectly, of the probability distribution of the
possible future state of the system, given its initial state and a time
lapse:

P ( 
x , t | x0 , t0 ) := P(X(t)  0 ) = x0 )
= x | X(t (2)

Equation (2) is the Markovian condition that assumes the future

state of the system relies exclusively on the present state (x0 , t0 ) and
the propensity function of each possible reaction. To be precise, to
get to state x at time t + dt for a dt small enough to ensure that
the probability of two reactions occurring in that time interval is
negligible, either the state at time t is also x, or the state at time t
is x − dr , and reaction r takes place during the interval [t, t + dt].
Thus, we have the following approximate equality (in which R is
the set of all reactions):
6 Álvaro Bustos et al.

x , t + dt | x0 , t0 )
P (

≈ x − dr , t | x0 , t0 ) P(reaction r happens in [t, t + dt])
P (
r∈R

x , t | x0 , t0 ) P(no reaction happens during [t, t + dt])

+ P (

≈ P (x − dr , t | x0 , t0 )a(r, x − dr ) dt
r∈R
 

x , t | x0 , t0 ) 1 −
+ P ( a(r, x) dt
r∈R

x , t | x0 , t0 )
From the last expression (Eq. (1)), moving the term P (
to the left side of the equality and dividing by dt, the following
identity is obtained as dt → 0:
d 
x , t | x0 , t0 ) =
P ( x − dr , t | x0 , t0 )a(r, x − dr )
[P (
dt r∈R

− P (
x , t | x0 , t0 )a(r, x)] (3)

Equation (3), commonly known as the Chemical Master Equa-

tion (CME), can be rigorously formalized from the laws of
probability and the theory of Markov processes [9], but for sim-
plicity we will use the informal derivation given above. Although
the previous equation is theoretically enough to determine the
probabilities involved in the simulation at any moment given
x , t | x0 , t0 )), determining an
an initial state (i.e., the function P (
explicit form of P analytically from the CME is usually extremely
hard. This difficulty is due to Eq. (3) being a system of coupled
differential equations with one function for each different state
vector , and thus it can potentially have infinite unknown functions.
Therefore, using this equation directly as a basis for a simulation
is extremely impractical for systems composed of many different
types of agents and/or with a large number of rules. However, it is
possible to construct accurate numerical Markov simulations that
follow the distribution given by the CME [10]. To accomplish
this and accurately simulate the future state of a system based on
information of the current state, only two questions need to be
answered:
• Which reaction will happens next?, and
• How much time will pass from now until it happens?
Thus, for an accurate simulation, we only need information about
the conditional probability distribution of the next reaction r and
expected time τ . So, we define the function p(r, τ | x, t) as follows:

p(r, τ | x, t) dτ ≈ P(the next reaction happens in the interval


[t + τ, t + τ + dτ ] and is of type r | X(t) = x)
 Rule-Based Models and Applications in Biology 7

If we assume the Markovian memoryless property, this probability

should be independent of the current time t; thus, the definition
can be simplified slightly by removing references to t:

p(r, τ | x, t) dτ ≈ P(no reactions during [0, τ ] and a reaction of


type r during [τ, τ + dτ ] | X(0) = x).

Assuming that every reaction r takes place independently of

all other reactions, the Markovian assumption tells us that the
expected time Tr until a reaction of type r is an exponential
variable of rate a(r, x) [17, chapters 6–7]. Thus, the time T =
minr∈R Tr until the next reaction is an exponential variable of rate
x ):= r∈R a(r, x), and it is independent of the reaction r chosen.
a0 (
Therefore, an explicit value for the probability density p can be
easily determined:

p(r, τ | x, t) = a(r, x) exp(−τ a0 (

x )). (4)

Equation (4) can be used to generate trajectories that follow

the desired distribution, since it implies that the probability of
choosing a given reaction r is a(r, x)/a0 (
x ) and it is independent
of the expected time T , we get the following simple algorithm for
generating valid trajectories given an initial state x0 :
 as x0 and the current time t to 0.
1. Initialize the state X
⎡ ⎤
1000 r1
 = t0 ) = ⎣ 900 ⎦ × A, B, C 2A +B −
→C
x0 = X(t r2
2C −
→ 2A + B + C
1200

2. Generate two random numbers p1 , p2 in [0, 1] (uniform distri-

bution), for example:

p1 = 0.18 and p2 = 0.67

x ) = r∈R a(r, x)

3. Determine the reactivity of the system as a0 (

and set δt as the value ln(1/p1 )/a0 (X). This ensures that
the random variable δt has an exponential distribution with
 For simplicity, each reaction occurs at the same
rate a0 (X).
frequency r = r1 = r2 = 1.0 s−1

x ) = r1 × A × (A − 1) × B + r2 × C × (C − 1)
a0 (
x ) = r1 × 1000 × (1000 − 1) × 900 + r2 × 1200
a0 (
× (1200 − 1)

δt = ln(1/0.18)/900,538,800 = 1.90 × 10−9 s = 1.90 μs

8 Álvaro Bustos et al.

4. Suppose the set of reactions is given by R = {r1 , . . . , rj , . . . , rn }.

The probability to choose any reaction r ∈ R is a(r, X)/a  0 (X). 
We choose the reaction rj testing the following inequality:


n−1 
a(rj , X) 
n 
a(rj , X)
≤ p2 <

a0 (X) 
a0 (X)
j =1 j =1

For example, given the two reactions r1 and r2 , we test the

following inequalities:


1 
a(rj , X)
0 ≤ 0.67 < = 0.9984

a0 (X)
j =1


2 
a(rj , X)
0.9984 ≤ 0.67 < = 1.0000

a0 (X)
j =1

5. Replace the old value of t by the new value t + δt and the old
 with X
value of X  + dr , where r is the reaction chosen in step
(4).
⎡
⎤ ⎡ ⎤ ⎡ ⎤
1000 −2 998
 = t0 + δt) = ⎣ 900 ⎦ + ⎣−1⎦ = ⎣ 899 ⎦
x1 = X(t
1200 1 1201

 and go back to step (2) or finish

6. Save the new values of t and X

if a0 (X) = 0.
This is a basic form of the SSA, readers interested in a more in-
depth analysis of the model may consult the review by Gillespie
in [10]. Common methods for the simulation of rule-based
models use adapted versions of this algorithm to generate accurate
simulations, each approach making certain assumptions and often
requiring a formal language to describe the models. Examples of
such SSA-based implementations are BioNetGen [3] and KaSim
[13], each with its own formal language (BNGL [7] and Kappa
[16], respectively).

2 Introduction to Ordinary Differential Equations Models

Another common approach to the study of the dynamic behavior of

complex systems employs ODEs or Partial Differential Equationss
(PDEs) based on the empirical law of mass action [12, 21]. This
law states that the rate of a chemical reaction is proportional to the
activity of each of its reactants. In order to simplify the model, it is
often assumed that such activity values match the concentrations
 Rule-Based Models and Applications in Biology 9

of each reactant. While this is generally not true, for elemental

reversible reactions with no intermediate steps, it is a reasonable
assumption and an acceptable approximation. For instance, given
an elemental reversible reaction such as the following:

A+BC

the rate at which the forward reaction A + B → C occurs is

proportional to the concentrations of A and B, with a similar
remark applying to the backward reaction. This simple reversible
equation prompts the following three ODEs systems as a candidate
for modeling its evolution or dynamic behavior over time:

d[A]
= −k1 [A][B] + k2 [C]
dt
d[B]
= −k1 [A][B] + k2 [C]
dt
d[C]
= k1 [A][B] − k2 [C]
dt

in which [X] stands for the concentration of the reactant X and

k1 and k2 are rate constants usually determined from experimental
data. The right-hand side of the equation represents that in the
forward reaction (A + B → C), one instance of A and one of B are
replaced by one of C, with the opposite happening for the reverse
reaction.
This small system of ODEs is usually nonlinear. The model has
a very simple structure, and allows both numerical and theoretical
analyses. For instance, equilibrium can be calculated assuming that
k1 [A][B] − k2 [C] = 0, which leads to Eq. (5):

k1 [C]
K= = (5)
k2 [A][B]

where K is called the equilibrium constant of the system and

does not depend directly on the concentrations of the reactive
substances but only on the rate constants k1 , k2 . K governs the
asymptotic behavior of the system as time goes to infinity [19, 20];
more precisely, a system of chemical reactions eventually reaches
a situation in which the concentration of each chemical involved
remains unchanged, with this value being determined by the
constant K [20, chapter 17]. This can be seen mathematically
by noticing that the system of equations above has a constant
solution whose value depends on K, and any other positive solution
converges to this value as t → ∞ [21].
However, for more complex reactions and systems with more
types of agents, the setup of the ODE system and the structure
of the resultant reactions become very difficult to simulate using
10 Álvaro Bustos et al.

this type of equation [10]. Chemical reactions such as electrolysis,

which involves two or more instances of the same reactant, intro-
duce higher-order terms that might induce unexpected and/or
difficult-to-explain behavior in numerical simulations. In addition,
non-elementary reactions have to be decomposed into a series
of elementary reactions, which can greatly increase the number
of terms and variables involved in the system. Thus, the ODE
approach becomes impractical very quickly in sufficiently complex
chemical systems. Another drawback of this approach is that low
concentrations or quantifications of agents can lead to unrealistic
simulations of the behavior of the system in the long term upon
extinction of these agents. This is particularly noticeable in small
systems comprised of only hundreds or thousands of agents.
Another characteristic of the ODE-based approach is that
it is purely deterministic. Given that in a real chemical system
there are random fluctuations and non-deterministic phenomena,
a deterministic model might not be able to fully represent all of the
possible outcomes of the system. As in the previous paragraph, it is
worth mentioning that random fluctuations usually have negligible
long-term influence in large systems with sufficiently high con-
centrations of every species in the system. However, they become
much more evident in systems with a lower number of compo-
nents. In such systems, there are potential alternative outcomes
(different from the average behavior simulated by deterministic
models) with large quantitative differences and non-negligible
probabilities. Hence, taking into account this non-deterministic
behavior becomes essential to understand small-scale systems [10].
Lastly, ODE-based models usually carry no spatial information,
as the medium is assumed homogeneous and well-stirred, with a
uniform distribution of all system components. Here, we describe
several biological systems in which those assumptions are invalid.
The most straightforward way to create models that take into
account spacial information is by replacing the concentration value
as a function of simulated time for each entity [X](t) by a spatial
density term ρX (t; x, y, z), which represents the density of X in
a small neighborhood of points in the area or volume comprised
by the model. Also, additional terms in the differential equations
above are required to model physical phenomena that may affect
density. For example, the chemical entities in the simulated system
are liquids capable of diffusion; a possible set of equations for the
reaction A + B → C could be defined as

∂ρA
= −ρA − kρA ρB
∂t
∂ρB
= −ρB − kρA ρB
∂t
∂ρC
= −ρC + kρA ρB
∂t
 Rule-Based Models and Applications in Biology 11

where each ρ term corresponds to the following sum of partial

derivatives (known as the Laplacian or Laplace operator):

∂ 2ρ ∂ 2ρ ∂ 2ρ
ρ = + +
∂x 2 ∂y 2 ∂z2

This conforms to the usual diffusion reaction from physics (as

stated in [6, Chapter 2]), ∂ρ/∂t + ρ = 0 (with constant diffusion
rates uniformly equal to 1), after adding the additional terms
brought by the law of mass action considering that for a very small
neighborhood of a point (x, y, z) the term [X] is proportional to
ρX and the chemical X may be assumed approximately homoge-
neous.

3 Parallel Implementation of Spatial κ

3.1 The κ In this section, we introduce a modified version of the SSA to

Algorithm allow more complex simulations in a variety of contexts beyond
the standard chemical applications. We do not go in depth into
the mathematical formalisms behind the modifications of the SSA
introduced here; these details may be consulted in publications
about the κ language such as the work of Danos et al. [4, 5].
The discussion below follows the theoretical framework setup by
Danos, with a schematic graphical notation whenever possible. For
the actual language and syntax used in standard implementations
such as KaSim, please consult the KaSim reference manual [14].
Nevertheless, the examples in this and the following sections can
be easily implemented in KaSim, which provides all of the standard
κ framework. Further examples that involve spatial information are
designed to be compatible with PISKaS [18], which is a spatially-
enhanced fork of KaSim.
The classical Gillespie’s algorithm treats every kind of chemical
compound (or, in general, a variation of an agent) as a separate
type, no matter how similar it may be to a previously existent
type of compound [5]. This becomes problematic when there is
a large amount of different compounds that are similar—but not
identical—as there is no way to express this similarity properly
in the classic SSA framework, even if these cannot participate in
the same reactions. Thus, this results in state-vectors with a large
number of entries and (usually) several almost-duplicate reactions
or rules involving small variants of the same compound, requir-
ing too many computational resources to simulate such systems.
Another problem is that the described SSA framework ignores the
internal structure of the compounds involved. This is a problem
when dealing with complex molecules such as proteins or DNA,
since their internal structure can severely influence the outcome of
a chemical process out of sheer geometrical positioning, let alone
12 Álvaro Bustos et al.

physical or chemical constraints caused by the size of the molecule.

Last, biological systems and other complex systems are naturally
compartmentalized (cellular compartments), a characteristic diffi-
cult to replicate into a model using an algorithm that assumes a
homogeneously mixed environment where all reactions take place.
The first two observations made above suggest that a modi-
fication of the data structure used to store the current status of
the system, as well as a change in the idea of what constitutes
an agent, might allow for a more flexible and robust framework.
The concerns about information regarding the fixed structure of a
compound suggest that an atom is probably a better model than
a molecule for the concept of agent. An atom interacts with other
atoms in several ways, the covalent bond being among the simplest
to understand conceptually: Each atom can form a finite number
of preestablished links of a specific type with one or more other
atoms, which in turn can also have links between themselves. For
instance, an oxygen atom can form two covalent bonds, or in other
words, it has two “open places” where other atoms can bind to,
while a hydrogen atom can form a single covalent bond. When two
hydrogen atoms bind to one oxygen atom by forming two covalent
bonds, a water molecule is formed. Similarly, chemical reactions
can be expressed as the formation or destruction of links between
reactants or agents.
This motivates storing the current state of the system as a site
graph [5]. This graph corresponds to a network in which the
nodes or agents have a specific structure that limits the kinds of
connections or bonds that can be formed. More specifically,
• Each agent has a type. Going with our chemical analogy, this
would correspond to the specific element (hydrogen, helium,
oxygen. . . ) of each atom.
• Each type has a set of sites associated with it. Every site has
a set of possible internal states. In our example, these sites
correspond to the places in the atom where covalent bonds
can be formed, while the internal states may correspond to
markers of phenomena like partial charges, or differentiators
between distinct types of chemical bonds.
• Each link between two nodes (agents) connects exactly one site
from one of the agents to one site of the other; reciprocally,
a site from an agent can be involved with at most one link
with another agent. In our chemical example, this means, for
instance, that each of the two “open positions” from an oxygen
atom can participate in only one covalent bond with another
atom and thus this atom can be bound to at most two other
atoms at once.
Reactions or rules can be also described via site graphs. A rule r is
expressed via a site graph Sr and a set of transformations Ar , which
 Rule-Based Models and Applications in Biology 13

corresponds to the addition or removal of edges between sites of

Sr , changing their internal states, or adding or removing agents
from Sr .
As a simple example, let us consider the reaction of electrolysis:

2H2 O → 2H2 + O2

To describe this reaction, only two types of agents are needed: H

and O, each one representing the type of atom. Agents of type H
have one site, h1 , while agents of type O have two sites o1 and o2 .
For our current purpose, neither of the sites has a specific internal
state. The reactants can be represented by a graph with six nodes
(agents), two of type O and the rest of type H; each of the four sites
oi , i = 1, 2 is linked to a single h1 site from one of the H agents.
Furthermore, the set of transformations to be applied to this graph
are as follows:
• Delete the four oi h1 links.
• Add a h1 h1 link to the two H agents corresponding to each
water molecule.
• Add two links, o1 o1 and o2 o2 , between the two O agents.
The effects of these operations on a set of agents can be observed
graphically in Fig. 1.
Observe that by the specific combinations of two types of
agents, we are able to describe three different species participating
in this reaction (H2 O, H2 , O2 ). With the same two agents, other
chemical species can be easily described. For example, ozone can
be described using three O agents and different internal states
to represent the hybridization of the chemical bonds involved.
Another example is hydrogen peroxide, which uses two H and two
O agents, with a pattern of links mimicking the chemical structure
of the molecule. By adding only a third type of agent with four sites
c1 , . . . , c4 , we can include carbon atoms in our model and thus
represent the whole set of hydrocarbon species and other related
types of compounds.

O O O
H H H H H H

H H H H H H
O O O

Fig. 1 Rearrangement of agents over the application of a rule corresponding to the reaction of hydrolysis,
2H2 O → 2H2 + O2
14 Álvaro Bustos et al.

The κ framework allows declaring certain internal states or

site links as “undefined,” which allows applying the same rule to
similar, but not identical, species. For example, the formation of
alcohols from hydrocarbons corresponds to a set of very similar
reactions, usually consisting of replacing a single H agent by a
two-agent subgraph corresponding to the radical −OH. Therefore,
specifying the whole structure of the hydrocarbon involved is
usually superfluous.

3.2 Non-chemical While the original motivation for the SSA comes from literal
Models in κ : A expression of chemical reactions, this framework can be used
Predator–Prey to model other types of systems where the agents involved do
Ecosystem not represent chemical units but instead more complex entities.
A simple example of this is the implementation of a predator–
prey model, where agents represent a predator species that may
consume other agents (prey). In this model, additional agents may
also be used to indicate the availability of limited resources, such as
plants or edible fruits for the sustenance of a herbivorous prey.
A simple system involving two species A, B (prey and predator,
respectively) can be modeled via a set of Lotka–Volterra equa-
tions (as seen in [15, chapter 7, section B]), which correspond to
the following system of differential equations:

dA
= αA − βAB
dt
dB
= γ AB − δB
dt

Here, α, β, γ , δ are nonnegative rate constants. The two terms of

the first equation are interpreted as follows: αA means that the
rate of growth of the prey species is proportional to the number
of extant members of the species (i.e., exponential growth); −βAB
represents the predation rate of members of A by the B species,
which assuming a homogeneous population is proportional to the
product AB. With respect to the two terms of the second equation,
these are interpreted as γ AB corresponds to the growth rate of the
predator species, which is proportional to the number of extant
members of B and A as well as to the number of available resources
or the amount of prey population; and the term −δB is the rate
of extinction of the predator, assumed to be proportional to the
current population of B.
Note that the rate of natural death of A is neglected (techni-
cally, it can be represented by a diminished value of the constant
α) as well as the dependence of A on other resources (for
instance, available plants for a herbivorous animal). Moreover, the
population density of both species is assumed to be constant. For
instance, sexual reproduction is not considered, no age groups are
taken into account (which makes this model inaccurate for predator
 Rule-Based Models and Applications in Biology 15

species that target young members of the prey species), and no

extinction of any of the involved species can be studied, since the
population densities are assumed to be constant.
The simple Lotka–Volterra model can be implemented as
a κ model, allowing inclusion of different parameters into the
model such as natural dead, variable population densities, sexual
reproduction, and age groups. In order to simplify notation, we
expressed the model as chemical equations with internal states
being represented via parentheses and linked sites via lines when-
ever necessary.
The rules of reproduction for A and extinction for B have a
very simple format:
r1
A−
→ 2A
r2
B−
→∅

The reproduction rule for B has A as a catalyst, as the frequency at

which reproduction of B occurs depends on the A population, as
B will attempt to reproduce more often if there are more resources
available, but this does not mean they should consume a member
of A every time they attempt to reproduce. Thus, the rule is
r3
A+B −
→ A + 2B

Finally, the predation rule has B as a catalyst; for it to occur, a

member of A needs to encounter a predator B. In this model,
“hunger” or similar states are not considered. Thus, the rule
appears as follows:
r4
A+B −
→B

The rates of each of those rules depend on the values of α, β, γ , δ

and they can be determined in the same way as if they were
chemical reactions. Note that this model does not take into account
internal states (e.g., hunger) or links between agents (e.g., two B
agents acting together to capture a prey). Next, we will discuss
possible improvements of the model by using internal states or links
to represent this type of situation.
One simple addition to this model would be the implemen-
tation of sexual reproduction. Of course, this will not apply to
every type of species, and its effects might be negligible in simple
ecological systems; however, for environments with large disparity
in sex distribution or acute sexual dimorphism, this approach might
provide an accurate model.
To implement sexual reproduction into the model, we can use
sites as a property of the agents. Sites are variables that can be used
to store a finite set of values or states in the form of qualitative or
quantitative descriptors. Thus, we can use a site g in each agent to
16 Álvaro Bustos et al.

represent the sex (e.g., ♀, ♂ for male and female, respectively, or

 for species with hermaphroditic individuals). Thus, the rules for
sexual reproduction are as follows:

A(♀) + A(♂) → A(♀) + A(♂) + A(♀)

A(♀) + A(♂) → A(♀) + A(♂) + A(♂)
A(?) + B(♀) + B(♂) → A(?) + B(♀) + B(♂) + B(♀)
A(?) + B(♀) + B(♂) → A(?) + B(♀) + B(♂) + B(♂)

Note the A(?) term in the left side of the predator reproduction
rules. As stated before, we allow for sites or links to be undefined so
a single rule can be applied to every combination of internal states
of A. In this case, what matters is that there are available resources
(i.e., prey) and not the specific sex of the prey animals present. The
A(?) term on the right-hand side means that the corresponding
term on the left-hand side remains untouched. These rules add
new agents of a specific type (A(♂), A(♀), B(♂), B(♀)) without
affecting the existing ones.
Age information or the stage of maturation of the agents can
also be useful to improve the Lotka–Volterra basic model. For
instance, we can suppose that predators more often capture young
or elderly animals of the prey species due to inexperience, physical
weakness, or illness. Similarly, only animals that have reached sexual
maturity can reproduce, and in some species elderly animals present
diminished fertility.
To incorporate this information into the model, we include an
additional internal state d, whose values correspond to the different
stages of development of each species, for instance {Dc , Dy , Da , De }
(child, young or adolescent, adult or sexually mature, elderly or
senescent, respectively). We need to define rules of growth that
make every agent transit through those internal states sequentially:

A(?, Dc ) → A(?, Dy )

For examples of sex- and age-segregated ecological models that

served as the inspiration for the set of rules shown here, see
Fundamentals of Mathematical Ecology, by Mark Kot [15].
Every rule introduced above could be reproduced in a rela-
tively simple way in the usual Gillespie’s framework. However, the
possibility to link agents through sites has not yet been covered in
the κ language in this section. As an example, we will consider the
formation of herds in both predator and prey species. A large group
of prey animals can fend off a lone predator, while a prey animal
can be more easily overwhelmed by a herd of social predators when
alone or in a small group.
A potential way to implement this would be to add a few sites
through which an agent can be linked to others of the same kind.
Those links can represent social relations in the herd, and we can
 Rule-Based Models and Applications in Biology 17

define rules to represent both herd protection and social hunting.

For instance, we can add a few “relation sites,” e.g., p ♀ , p ♂ , c, m
(mother, father, child, mate) to represent a monogamous species
with just one offspring per reproduction event and define the
following state rules:

m
m
B(♀, Da ) B(♂, Da ) → B(♀, Da ) B(♀, Dc ) B(♂, Da )
c,p♀ p♂ ,c

s,t
Here, the notation B B means site s from agent B is linked
to site t of another agent B. Thus, this rule reads as follows: two
agents who are in the “adult” stage of development and are a
mating couple (there is a link between the m sites of both agents)
engender a third agent (in this case, female) and the c site of each
parent agent gets linked to the respective p site of the new B(♀, Dc )
agent (the former two agents get marked as parents to the new
child agent).

3.3 Spatial Usually, the setup for the simulation algorithm of the κ framework
Information in κ in the standard implementations of κ such as KaSim [5, 14] makes
the same physical assumptions as the standard SSA, in particular
that the medium is homogeneous and well-stirred, which means
that the agents are uniformly distributed in the environment.
This simplifies the model defining the probability of two agents
interacting as proportional to their respective population. While
this assumption is valid for certain systems, e.g., chemical reactions
in gases, it might not be applicable to systems that are not
homogeneous or have spatial dependences. For instance, the cell
membrane provides different chemical and physical properties to
the intracellular and extracellular medium. Moreover, the transfer
of certain substrates from one medium to the other is in itself a phe-
nomenon of interest, which is entirely ignored by the κ framework.
Thus, incorporating spatial information to an implementation of
the κ language allows for more realistic models. However, the
assumption of homogeneity cannot be completely eliminated, as
we usually care only about large-scale tendencies and not individual
agent behavior.
In the cell membrane system, there are two different and clearly
defined media. Interactions between them consist of transportation
of certain agents from one environment to the other through the
membrane. We assume that both environments or cellular com-
partments are homogeneous; hence, the probability of a certain
agent approaching the membrane depends on the quantity of that
agent, which can be represented using rules.
18 Álvaro Bustos et al.

One way to implement such system of compartments is simply

to add a new site w to every agent to represent each medium, which
can have two states i and o corresponding to inside and outside
the cell. Each rule becomes a set of rules, one per compartment,
to ensure that agents only interact with other agents in the same
compartment:

A(i) + B(i) → C(i)

A + B → C becomes
A(o) + B(o) → C(o)

The transport rules change the states of agents from one com-
partment to another. In this case, we can represent this via rules
like A(i) → A(o) and A(o) → A(i), each with a certain rate.
In this way, to simulate osmosis through equal volumes, each of
those rules should have equal rate, such that the compartment with
higher concentration has a higher rate of transportation.
For a more complete system with many more compartments,
we can separate the cell space into a series of subspaces that we
assume approximately homogeneous. This is analogous to the
Riemann sums method to compute an integral:

1 
n−1
f (k/n)
f (x) dx ≈
0 k=0
n

To recall, this method allows to approximate the area under a

curve given by the function f as a sum of the areas of small
rectangles [2]. Here, we divide the interval [0, 1] into n equal
subintervals (subspaces) of length 1/n, and we assume that the
value of f in [k/n, (k + 1)/n] is approximately f (k/n), i.e, this
assumption does not significantly affect the value of f and it is
similar to the “approximately homogeneous” assumption from
above. For Riemann summations, a large value of n, and, thus,
1
smaller subintervals, gives a better estimation of 0 f (x) dx, at
least when f is a continuous function. In a similar way, we can
suppose that with smaller subspaces we should reach a better
approximation.
If the space of agents has some kind of geometrical properties,
we can represent them via the transport rules: Compartments that
are geometrically adjacent should have a higher rate of transfer
reactions. For instance, we could represent a cell as two different
compartments, nucleus and cytoplasm (Fig. 2), interconnected by
several transport mechanisms, or build a more complex system
where the nucleus is represented as a central compartment sur-
rounded by several other compartments representing subspaces of
the cytoplasm (Fig. 3).
To represent the geometry of the system, here we assume that
there are only transport rules between adjacent compartments,
 Rule-Based Models and Applications in Biology 19

Cytoplasm

Nucleus

Nucleus Cytoplasm

Fig. 2 Internal representation and interpretation of a two-compartment model

for a cell. Note that the geometry of the cell is not taken into account and thus
the cytoplasm and the nucleus are taken as entirely homogeneous

Fig. 3 Potential compartment arrangements to represent the geometry of a

cell. (Left) A 2D arrangement of nine compartments with eight representing
the cytoplasm and the central one representing the nucleus. The arrows
represent possible transport rules. (Right) A potential 3D version of the former
2D arrangement. The marked (central) compartment corresponds to the nucleus,
while the other 26 represent the cytoplasm; there are transport rules among
compartments that share a face

automatically considering that the rate of any transport rule

between nonadjacent compartments is zero.
As stated before, this does not require a special implementation
of κ and can be done in the usual implementations such as KaSim
by using a special site whose states correspond to the different
compartments. However, since we need a copy of a rule for each
compartment, this may result in several redundant rules that may
severely impact execution time. Thus, an implementation allowing
explicit declaration of compartments can reduce the total number
of rules needed and improve the performance of the simulation.
In what follows, we will discuss one such implementation, namely
Parallel Implementation of a Spatial Kappa Simulator (PISKaS)
[18], which is a forked branch of KaSim. PISKaS focuses on
independent, parallel simulation of each compartment; however,
most remarks below should apply to any spatial κ implementation.

3.4 In the following section, we will introduce a model of population

Metapopulation dynamics that attempts to reproduce the experimental results
Dynamics in a obtained in the work of Holyoak and Lawler [11]. The study
Predator–Prey subject here was the evolution of the population of a set of predator
Model with Explicit and prey species in an environment that allows spatial migration.
Spatial Information The goal is to verify whether a prey species who is prone to
extinction by predation can survive in a medium that allows for
20 Álvaro Bustos et al.

migration. The end result of the simulation appears to conform to

the experimental results; further details of an implementation of
this simulation can be consulted in [8].
The experiment in [11] studied two species of microor-
ganisms, Didinium nasutum (predator) and Colpidium striatum
(prey), which inhabit an environment consisting of several bottles
(compartments) linked by four-way connectors in a specific config-
uration. Given that the prey species shows logistic behavior (i.e.,
short-term exponential growth, which eventually is stunted due to
lack of resources, resulting in a stable population) in the absence
of predators, it becomes natural to suggest a Lotka–Volterra model
to represent the interaction between those two species, with rules
similar to the ones introduced in Sect. 3.2.
This model incorporates transport rules to represent microor-
ganisms from both species moving through the different bot-
tles. A coarse geometry has to be introduced, representing the
spatial arrangement of the system of bottles and connections.
Each bottle can be assumed to be a homogeneous medium and,
thus, can be taken as a compartment. In the original exper-
iment, the configurations consisted of square grids of bottles
joined by four-way connectors that linked each bottle to the
ones adjacent horizontally, vertically, and diagonally, as seen in
Fig. 4.

Fig. 4 Representation of the bottle arrangement and the four-way connections

linking each bottle to its neighbors
Other documents randomly have
different content
juice of one lemon, a saltspoonful of salt and a tablespoonful of
butter. Boil ten minutes.
MENUS AND INDEXES
 CHAPTER XXII.
PAPER BAG MENUS FOR WINTER.

BREAKFAST NO. 1.

Grapefruit
Cereal
Sweetbreads with Bacon (Paper-bagged)
Scones (Paper-bagged)
Coffee.

BREAKFAST NO. 2.

Oranges
Cereal
Spindled Oysters with Bacon (Paper-bagged)
Water Cress
Warmed over Rolls (Paper-bagged)
Coffee.

BREAKFAST NO. 3.

Baked Apples (Paper-bagged)

Beefsteak Leftovers (Paper-bagged)
Sweet Potatoes Southern Style (in paper-bag)
Scones (Paper-bagged)
Coffee.
 LUNCHEON OR SUPPER NO. 1.

Chicken Croquettes (Bagged)

Olives Pickles
Hot Biscuit (Bagged)
Gingerbread (Bagged)
Cheese
Tea.

LUNCHEON OR SUPPER NO. 2.

Oyster Bundles (Bagged)

Baked Potatoes (Bagged)
Celery Olives
Pork Cake (Bagged)
Baked Quinces (Bagged)
Cocoa.

LUNCHEON OR SUPPER NO. 3.

Mock Fried Oysters (Bagged)

Pickles Celery
Sally Lunn (Bagged)
Sponge Cake (Bagged)
Baked Apples
Tea.

DINNER NO. 1.

Grapefruit with Maraschino Cherries

Olives Pickles
Smelts Milanaise (Bagged)
Roast Chicken (Bagged) Baked Potatoes (Bagged)
Currant or Cranberry Jelly (Bagged)
Baked Onions (Bagged) Lettuce Salad
Plum Pudding (Bagged) Hard Sauce
Demi-Tasse.

DINNER NO. 2.

Grilled Sardines on Crackers (Bagged)

Ripe Olives Celery Salted Almonds (Cooked in Bag)
Ducks (Roasted in Bag)
Candied Sweet Potatoes Southern Style (in Bag)
Cranberry Molds, Biscuit (Bagged)
Baked Apples Stuffed with Nuts (Bagged)
Served with Cream
Gingerbread (Bagged)
Tea.

DINNER NO. 3.

Anchovy Canapés (Bagged)

Olives Celery
Roast Veal (Bagged)
Baked Potatoes (Bagged)
Spinach (Paper Bagged)
Endive and Roquefort Cheese Salad
Cheese Straws (Paper-bagged)
Mince Pie (Paper Bagged)
Black Coffee.

PAPER BAG MENUS FOR SPRING.

 BREAKFAST NO. 1.

Baked Rhubarb and Raisins (Paper-bagged)

Cereal
Omelette (Paper-bagged)
Crisped Sweet Potatoes (Paper-bagged)
Rolls (Reheated in bag)
Coffee.

BREAKFAST NO. 2.

Strawberries au Naturel
Cereal
Eggs in Cocottes (Paper-bagged)
Scones (Paper-bagged)
Coffee.

BREAKFAST NO. 3.

Baked Prunes (Paper-bagged)

Cereal
Sweetbreads (Bagged) Water Cress
Baking Powder Biscuit (Bagged)
Coffee.

LUNCHEON OR SUPPER NO. 1.

Rhubarb Short Cake (Paper-bagged)

Cold Veal Loaf (Paper-bagged)
Chocolate Cake (Bagged)
Tea.
 LUNCHEON OR SUPPER NO. 2.

Crab Meat au Gratin (Paper-bagged)

Biscuit (Paper-bagged)
Mrs. Kelder's Loaf Cake (Bagged)
Strawberries
Cocoa.

LUNCHEON OR SUPPER NO. 3.

Chicken Croquettes (Paper-bagged)

Biscuit (Bagged)
Pickles Olives
Good Friday Cake (Paper-bagged)
Custard
Tea.

DINNER NO. 1.

Caviare Canapés (Bagged)

Salted Nuts (Bagged) Olives
Roast Leg of Lamb (Bagged) Mint Sauce
Baked Potatoes (Bagged)
Stuffed Baked Onions (Bagged)
Rhubarb Pie (Bagged)
Coffee.

DINNER NO. 2.

Bouchees of Sardines (Bagged)

Deviled Almonds (Bagged) Radishes
Breast of Lamb with Tomato Sauce (Bagged)
Parsnips (Bagged)
Baked Potatoes without their Jackets (Bagged)
Lettuce Salad
Rhubarb Short Cake (Bagged)
Black Coffee.

DINNER NO. 3.

Strawberries au Naturel on Orange Slices

Mussels au Gratin (Bagged)
Irish Stew (Bagged)
Scalloped Tomatoes (Bagged)
Lettuce Salad
Lemon Pie
Coffee.

PAPER BAG MENUS FOR SUMMER.

BREAKFAST NO. 1.

Raspberries
Cereal
Creamed Mushrooms (Bagged)
Toast (Bagged)
Coffee.

BREAKFAST NO. 2.

Blackberries with Cream

Moulded Cereal
Crisped Bacon and Liver (Bagged)
Rolls (Bagged) Radishes
Coffee.
 BREAKFAST NO. 3.

Cantaloupe
Moulded Farina
Corn Fritters (Bagged)
Baked Egg In Tomato Cases (Bagged)
Scones (Bagged)
Coffee.

LUNCHEON OR SUPPER NO. 1.

Peach Puree
Potato Salad
Veal Loaf (Bag-cooked)
Raspberry Short Cake (Bag-cooked) with Cream
Iced Tea.

LUNCHEON OR SUPPER NO. 2.

Cold Game Pie (Cooked in Bag)

Hot Biscuit (Cooked in Bag)
Oatmeal Crisps (Cooked in Bag)
Blackberries
Iced Tea.

LUNCHEON OR SUPPER NO. 3.

Stuffed Tomatoes with Cream (Bag-cooked)

Baked Lamb, Sweetbreads (Bag-cooked)
Bread and Butter
Lettuce Salad
Raspberries Potato Caramel Cake (Bag-cooked)
 Iced Tea.

DINNER NO. 1.

Canteloupes
Radishes Olives
Lamb Chops (Bagged) Mint Jelly
Green Peas (Bagged)
String Bean Salad
Lemon Ice

DINNER NO. 2.

Sardines and Lemon

Olives Radishes
Saute of Chicken with Mushrooms (Bagged)
Sweet Potatoes en Brochette (Paper-bagged)
Sliced Tomatoes with French Dressing
Fruit Syllabub
Potato Chocolate Cake (Baked in Bag)
Iced Tea.

DINNER NO. 3.

Watermelon
Roast Lamb (Paper-bagged) Mint Sauce, Currant Jelly
New Potatoes (Bagged) Parsley Sauce
Oriental String Beans (Paper-bagged)
Cucumbers (Dressed with oil and vinegar)
Neufchatel Cheese and Wafers
Lemon Ice Chocolate Wafers (Bag-cooked)
Iced Tea with Lemon.
PAPER BAG MENUS FOR AUTUMN.
BREAKFAST NO. 1.

Peaches and Cream

Cereal
Fried Tomatoes (Paper-bagged) Cream Gravy
Blueberry Biscuit (Paper-bagged)
Coffee.

BREAKFAST NO. 2.

Baked Apples (Bagged-cooked) with Cream

Cereal
Eggs Baked in Tomatoes (Paper-bagged)
Baked Potatoes (Bagged)
Biscuit (Bagged)
Coffee.

BREAKFAST NO. 3.

Canteloupe
Ham with Apples (Bagged)
Sweet Potatoes (Bagged)
Corn Meal Gems (Bag-cooked)
Coffee.

LUNCHEON OR SUPPER NO. 1.

Cold Roast Chicken (Paper-bagged)

Baked Potatoes (Bagged)
Tomatoes with Mayonnaise
 Bread and Butter Folds
Baked Sweet Apples with Cream (Bagged)
Chocolate Cake (Bagged)
Tea.

LUNCHEON OR SUPPER NO. 2.

Corn Patties (Bagged)

Scalloped Potatoes (Bagged)
Olives Pickles
Farmer's Fruit Cake (Bagged)
Baked Quinces
Tea.

LUNCHEON OR SUPPER NO. 3.

Baked Potatoes en Surprise (Bagged)

Chicken Croquettes (Paper-bagged)
Sliced Tomatoes with French Dressing
Baked Apples with Nuts (Bagged)
Gingerbread (Bagged)
Tea.

DINNER MENU NO. 1.

Canteloupe
Caviare Canapés (Cooked in Bag)
Sauer Braten with Carrots and Onions (Bagged)
Baked Potatoes (Bagged)
Lima Beans (Bagged)
Sliced Tomatoes
Peach Short Cake (Paper-bagged)
Coffee.
 DINNER MENU NO. 2.

Caviare Canapés (Cooked in Bag)

Deviled Chestnuts (Paper-bagged)
Roast Pork (Bagged)
Sweet Potatoes (Bagged)
Baked Egg Plant (Bagged)
Cucumbers
Apple Pie (Paper-bagged) with Cream Cheese
Coffee.

DINNER MENU NO. 3.

Grapes and Peaches

Cream of Chestnut Soup with Croutons (Cooked in Bag)
Roast Duck (Bagged) Spiced Grapes
Sweet Potatoes (Bagged)
Baked Tomatoes (Bagged)
Grape Pie (Baked In Bag)
Coffee.
 CHAPTER XXIII.
A FEW OF THE EASIEST DISHES FOR
BEGINNERS

Baked Potatoes in their Jackets Page 96

Baked Potatoes without Jackets " 96
Bacon and Apples " 70
Sausage and Apples " 72
Bacon and Bananas " 70
Sausage with Tomatoes " 73
Roast Loin of Pork " 72
Hot Cheese Canapés " 20
Caviare Canapés " 20
Cheese and Cracker Canapés " 20
Cracker Crisps " 21
Roast Clams " 26
Lobster in Shells " 29
Baked Blue Fish " 31
Filets of Flounder " 34
Lamb Chops " 67
Roast Leg of Lamb " 69
Roast Chicken " 50
Vealettes " 76
Baked Onions " 94
Sweet Potatoes and Bacon " 97
Spinach " 98
Peas " 94
Turnips " 99
Baking Powder Biscuits " 101
Baked Apples " 112
Cinnamon Apples " 113
Apple Dumplings " 112
Baked Pears " 115
Mrs. Kelder's Loaf Cake " 107
Oatmeal Cakes " 109
Pork Cake " 109
Mince Turnovers " 122
Individual Apple Tart " 120
 INDEX
APPETIZERS AND RELISHES: Page
Bouchee Cases 18
Bonne Bouchee 19
Bouchees of Caviare, Olives and Mayonnaise 19
Bouchees of Sardines 19
Bouchees of Sausage or Tongue 19
Canapés, The Making of 19
Anchovy Canapés 20
Caviare Canapés 20
Hot Cheese Canapés 20
Cheese and Crackers Canapés 20
Cheese Toast Sandwiches 20
Cracker Crisps 21
Deviled Crackers 21
Diables à Cheval 21
Nut Appetizers 21
Salted Almonds 21
Deviled Almonds 22
Roasted Chestnuts 22
Salted Chestnuts 22
Deviled Chestnuts 22

BEEF:
Bullock's Heart 61
Stewed Bullock's Heart 61
Filet of Beef 61
Hamburg Steak 62
 Pot Roast 63
Rib Roast of Beef 63
Roast Round of Beef in Paper Bag 64
Sauer Braten 64
Beef Steak 65
Toledo Beef Steak 65
Stuffed Roast Beef or "Mock Duck" 65

CAKES:
Cheese Cakes 104
Cinnamon Cake 105
English Fairy Cakes 105
Fruit Cookies 106
Mrs. Godfrey's Soft Ginger Bread 106
Good Friday Cake 106
German Honey Cakes 107
Pecan Kisses 107
Mrs. Kelder's Loaf Cake 107
Hickory Nut Macaroons 108
Walnut Macaroons 108
Maple Sugar Cake 108
Molasses Coffee Cake 108
Nut Cake 108
Oatmeal Cakes 109
German Peach Cake 109
Pork Cake 109
Potato Chocolate Cake 110
Potato Caramel Cake 110
Auburn Pound Cake 111
Raisin Nut Cake 111
Sour Cream Cake 111
CHEESE AND EGG DISHES:
Cheese Balls with Tomato Sauce 87
Cheese Fritters to Serve with Salad Course 87
Pepper Cheese 87
Cheese Ramekins 88
Cheese and Eggs 88
Baked Eggs 88
Baked Eggs with Cheese 88
A Paper Bag Omelette 88
Cheese Omelette 89
Swiss Eggs 89
Eggs in Tomato Cups 89

FISH (also see Shell Fish):

Filet of Bass 31
Baked Blue Fish 31
Bloaters, A Breakfast Dish of 31
Cat Fish 32
Codfish Cones 32
Codfish à la Crême 32
Eels, Paper Bagged 33
Flounder à la Meuniére 33
Filets of Flounder 34
Finnan Haddie 34
Fish Cakes 34
New England Fish Pie 35
Fish Soufflé 35
Planked Fish Bag Cooked 36
Halibut à la Poulette 37
Herring au Gratin 37
Herrings with Herbs 37
 Kedgeree 37
Kippered Mackerel with Fine Herbs 38
Salmon Loaf 38
Scalloped Salmon 38
Salmon Soufflé 39
Baked Shad 39
Shad Roe 39
Smelts 40
Bagged Weak Fish 40
White Fish Planked 41

FISH SAUCE (also see Sauces and Gravies):

Anchovy Sauce 42
Quick Bearnaise Sauce 42
Bearnaise Sauce 42
Brown Sauce 43
Curry Sauce 43
Egg Sauce 43
Sauce Hollandaise 43
Egg Sauce Made from the Hollandaise 44
Lobster Sauce 44
Maitre d'Hotel Butter 44
Sauce for Broiled Shad à la Murray 45
Parsley Butter 45
Sauce Tartare 45

FRUITS:
Baked Apples 112
Baked Apple Dumplings 112
Cold Baked Apples with Rum 112
Cinnamon Apples 113
 Apples Stuffed with Figs 113
Baked Apples and Nuts 113
Raisin Apples 114
Baked Apple Sauce 114
Baked Bananas 114
Stuffed Dates 114
Baked Gooseberries 114
Baked Peaches 114
Baked Pears 115
Baked Plums 115
Baked Quinces 115
Baked Raisins 115
Chestnut Patties 115

GAME (see Poultry and Game):

LAMB AND MUTTON:
Breast of Lamb with Tomato Sauce 67
Lamb Chops 67
Lamb or Mutton Cutlets with Tomatoes 67
Lamb Fry 68
Lamb's Kidney 68
Leg of Mutton Cooked in Cider 68
Mutton Chops and Sausage 68
Ragout of Lamb 68
Roast Leg of Lamb 69
A Genuine Irish Stew 69

PASTRY:
Plain Pie Crust 116
Apple Pie 117
Deep Apple Pie with Cream Cheese 117
 Cranberry Pie 118
Cranberry and Raisin Pie 118
Lemon Pie 118
Mince Pie 118
Mock Mince Pie 119
Pecan Pie with One Crust 119
Real Old-Fashioned Pumpkin Pie 119
Individual English Apple Tart 120
Colonial Pumpkin Tartlets 121
Apple and Cheese Turnovers 121
Apricot or Plum Jam Turnovers 122
Mince Turnovers 122

PORK IN VARIED FORMS:

Bacon and Apples 70
Bacon and Bananas 70
Bacon and Calf's Liver 70
Baked Pork Chops 70
Pork Chops and Sweet Potatoes 70
Ham and Scalloped Potatoes 71
Ham, Spinach and Lamb Chops 71
Stuffed Fresh Ham or Shoulder 72
Roast Loin of Pork 72
Roast Spare-Rib 72
Baked Sausage with Apples 72
Baked Sausage and Potato 72
Baked Sausage with Toast 73
Baked Sausage with Tomatoes 73
Tenderloin of Pork 73

POULTRY AND GAME:

Capon 47
Chicken with Parsnips 48
Chicken à la Baltimore 48
Chicken Croquettes 48
Paper Bagged Chicken 49
Chicken Pie 49
Paste for Chicken Pie 50
Chicken Rissoles 50
Roast Chicken 50
Saute of Chicken with Mushrooms 50
Smothered Chicken 51
Ducks with Banana Dressing 51
Canvas Backs 51
Chicken, Italian Style 52
Roast Wild Duck 52
Roast Wild Duck, Ohio Style 53
Frogs' Legs 53
Paper Bag Roast Goose 53
Sage and Potato Stuffing 54
Bag Roasted Young Guinea Fowl 54
Bag Broiled Young Guinea Hen 55
Quail 55
Stuffed Quail 56
Rabbit Cookery 56
Barbecued Rabbit 56
Roast Rabbit 57
Stewed Rabbit 57
Reed Birds 58
Squab 58
Barbecued Squirrel, (Southern Style) 58
Turkey à la Bonham 59
Venison 60
Venison Steak 60
PUDDINGS AND PUDDING SAUCES:
Almond Pudding 125
Apple and Fig Pudding 125
Banana Pudding 125
Farmer's Plum Pudding 126
Peach Betty 126
Peach Cobbler 126
Peach Roly-Poly 127
Plum Roly-Poly 127
Rye Bread Pudding 127
Tapioca Apple Pudding 128
A White Plum Pudding 128
Caramel Sauce 128
Cornstarch Pudding Sauce 129
Cream Sauce 129
Cream Sauce à la Hotel Astor 129
Delicious Fruit Sauce for Plum Pudding 129
Hard Sauce for Plum Pudding 129
Molasses Sauce 130

RECOOKED DISHES:
Beef Steak Left Overs 83
Chicken Croquettes 83
Mock Fried Oysters 84
Turkey Croquettes 84
Edinboro Hot Pot 84
Individual Meat Pies 85
English Pasties 85
Olla Podrida Pie 85
Oyster Bundles 86
SAUCES AND GRAVIES:
Bignon's Sauce 78
Bread Sauce 78
Brown Sauce 78
Celery Sauce 79
Currant Jelly Sauce 79
Curry Sauce 79
Hollandaise Sauce 79
Horseradish Sauce 80
Maitre d'Hotel Butter 80
Mexican Sauce 80
Mint Sauce for Roast Lamb 80
French Mustard Sauce, Creole Style 81
Mustard Sauce for Cold Meat 81
Onion Sauce 81
Spanish Sauce 81
Thick Tomato Sauce 82
Sauce Tartare 82

SHELL FISH:
Clam Pies 26
Roast Clams 26
Crabs, Soft and Hard 26
Creamed Crabs 27
Crabs Deviled à la William Penn 27
Crab Meat au Gratin 27
Crab Flakes au Gratin 28
Lobster Chops 28
Coquilles of Lobster 28
Lobster in Shells 29
 Mussels au Gratin 29
Boxed Oysters (Virginia Style) 29
Spindled Oysters and Bacon 30

SHORT CAKES:
Banana Short Cakes 123
Peach Short Cake 123
Rhubarb Short Cake 124
Old-Fashioned Strawberry Short Cake 124

SOUP ACCESSORIES:
Bread Sticks 23
Croutons Toasted 23
Crisped Crackers 23
Egg Balls 23
Forcemeat Balls, or Quenelles 24

VEAL:
Baked Calf's Liver 74
Calves' Brains in Tempting but Inexpensive Ways 74
Breaded Brains 74
Sweetbreads 75
Baked Sweetbreads 75
Sweetbreads with Bacon 75
Larded Sweetbreads 75
Sweetbreads Straight 76
Vealettes 76
Veal Loaf 76
Shoulder of Veal Stuffed and Braised 77
VEGETABLES:
Asparagus 90
Asparagus with Cheese 90
Lima Beans 90
String Beans, Oriental Style 91
Boston Baked Bean Cakes 91
Bean Croquettes 91
German Cabbage 92
Cabbage Hot Slaw 92
Carrots 92
Carrot Saute 92
Dolmas 99
Stuffed Eggplant 93
Lentil Cutlets 93
Mushrooms 93
Baked Onions 94
Stuffed Baked Onions 94
Onions with Cheese 94
Parsnips 94
Green Peas 94
Stuffed Peppers 95
Peppers with Cream Fish 96
Baked Irish Potatoes 96
Baked Potatoes without their Coats or Jackets 96
Potatoes en Surprise 96
Potatoes Farci 97
Sauer Kraut 97
Waldorf Sauer Kraut 97
Sweet Potatoes and Bacon 97
Sweet Potato Straws 98
Sweet Potato en Brochette 98
Spinach 98
 Summer Squash in Butter 98
Stuffed Summer Squash 98
Stuffed Tomatoes with Cream 98
Turnips 99
Turnip Balls 99
Stuffed Vine Leaves or Dolmas 99

WARM BREADS, BISCUITS, MUFFINS, ETC.:

Baking Powder Bread 101
Bannocks 101
Baking Powder Biscuits 101
Egg Biscuits 102
Maple Biscuits 102
Nut Biscuits 102
Raisin Biscuits 103
Hot Cross Buns 103
Warmed Over Breads 103
 OVAL WOOD
Cookery Dishes
Should
be
Used in
All
Paper
Bag
Cooking
They are as Important as
the Bags
Because they conserve all the delicate meat and vegetable juices,
adding a savory flavor to everything cooked in them.
With our Cookery Dishes you can give to all meats the delicious taste
which has heretofore been secured only by planking steaks and fish.
The sweet wood—we use sugar-maple only—is always fresh, giving
an effect that cannot be maintained permanently by the ordinary
plank.
Everything that can be cooked in a paper bag tastes better if you
use our Cookery Dishes also.
ASK YOUR DEALER ABOUT THEM
They are packed in cartons suitable for all purposes, assuring the
delivery of clean and sanitary dishes in your kitchen.
THE OVAL WOOD DISH COMPANY
Delta, Ohio
127 Franklin St., New York 436 Gravier St., New Orleans
Manufacturers of "O.W.D." Butter Dishes, Picnic Plates, and Clothes Pins

Refined
Vegetable Oil
Is recommended by physicians and culinary experts in place of
butter and animal fats for all cooking; it is more healthful and
economical.

Wesson Snowdrift Oil

The Best Refined Vegetable Oil
Is Unexcelled for
Greasing Paper Bags
 Y ou can buy many different kinds of
vegetable oils, but you can't get anything
equal to Wesson Snowdrift Oil. It is refined by
the Wesson process (the only process yet
discovered for properly refining vegetable oils)
and we control that process. No other
manufacturer can use it. ¶Wesson Snowdrift
Oil has just the right smoothness and
consistency to make rich and delicious salad
dressings.
AT ALL GROCERS
On request, we will mail you our Wesson
Snowdrift Oil book of 150 recipes. Please
mention your grocer's name.
———————————
The Southern Cotton Oil Company
Dept. B
24 Broad Street, New York, N. Y.
Enlarge.
Savannah Chicago New Orleans
San Francisco
 The Cookery Bag Clip
is the only s u c c e s s f u l device for effectually closing Paper
Cookery Bags

The projecting lips permit the clips to slip on to the

bags easily; the free ends projecting outwardly
prevent the clips slipping off the bag when in use.

Made by
THE OAKVILLE COMPANY
Waterbury, Conn.
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com