SimPed

SimPed simulates haplotype and genotype data for pedigrees of arbitrary size and structure. Founders are drawn from user-supplied haplotype frequencies (haplotype mode) or allele frequencies under Hardy-Weinberg equilibrium (genotype mode); non-founder haplotypes are passed down meiotically with recombination governed by user-supplied inter-marker recombination fractions or map distances. The two modes can be interleaved within a single simulation. SimPed handles diallelic and multiallelic markers and has been exercised on simulations with more than 20,000 loci.

This repository hosts a modified version of the original 2005 SimPed source that compiles cleanly with current C compilers (clang or gcc, -std=c99 or later) on macOS, Linux, and Windows. The algorithm and output format are unchanged; output is byte-identical to the original precompiled binaries on all bundled examples.

Reference. Leal SM, Yan K, Müller-Myhsok B. SimPed: a simulation program to generate haplotype and genotype data for pedigree structures. Human Heredity 60(2):119–122 (2005). PMID: 16224187.

---

Contents

1. Installation
2. Running SimPed
3. Input files
   • Pedigree file
   • Parameter file
4. Worked examples
5. Output file
6. Algorithm
7. Compile-time limits
8. Changes from the upstream source
9. License and contact

---

Installation

Build from source (recommended)

A C99 (or later) compiler and make are the only requirements. On macOS the Xcode Command Line Tools provide both (xcode-select --install); on Linux, any distribution package containing gcc or clang plus make will do.

make                              # builds ./simped
make debug                        # debug build (-g -O0)
make check                        # builds and runs two example inputs
make install                      # installs to /usr/local/bin
make install PREFIX=$HOME/.local  # user-local install, no sudo

Use a precompiled binary (not maintained; use as-is)

The executable/ directory contains the original 2005/2007 distribution binaries, organized by platform.

Platform	Binary	Notes
Linux	executable/linux/simped	ELF 32-bit Intel. On 64-bit Linux the runtime libraries libc6-i386 (Debian/Ubuntu) or glibc.i686 (Red Hat/Fedora) must be installed
macOS	executable/macos/simped	Mach-O x86_64. Runs on Apple Silicon under Rosetta 2. Native build via make is preferred
Windows	executable/windows/simped.exe + cygwin1.dll	Both files must reside in the same directory

Copy the binary onto a directory in your PATH to invoke it as simped from any location.

---

Running SimPed

simped <parameter_file>

The parameter file declares the input pedigree filename and the output filename on its first line. Both paths are resolved relative to the current working directory.

$ cd examples && ../simped input1.dat
  ********************************************************
  *                                                      *
  *               Program   SimPed                       *
  *                                                      *
  ********************************************************

Usage: simped parameter_file

Constants in effect:
  Maximum number of pedigrees                   100
  Maximum number of individuals                 3000
  Maximum number of loci                        10000
  Maximum number of haplotypes                  1000
  Maximum number of alleles at a locus          20

Input pedigree file: pedin1.pre
Output file: pedfile1.pre

---

Input files

SimPed reads two whitespace-separated text files: a pedigree file in standard LINKAGE format and a parameter file controlling the simulation.

Pedigree file

LINKAGE-style. Each line is one individual.

Column	Meaning
1	Pedigree identifier
2	Individual ID
3	Father's ID (0 if founder)
4	Mother's ID (0 if founder)
5	Sex (1 = male, 2 = female)
6–N	Affection status and/or quantitative trait columns (may be omitted)
N+1	Marker mask: which marker genotypes to emit for this individual

The marker mask resolves as follows:

• Single 0: no genotypes emitted, all markers blanked to 0 0.
• Single 1: all markers emitted.
• A string of 0/1 of length equal to the total number of markers: per-marker mask, where 1 emits the simulated genotype and 0 blanks it.

Pedigree and individual IDs may be numeric, alphabetic, or mixed. Multiple pedigrees may share one file, delimited by a change in the pedigree identifier column.

Parameter file

The header is seven fixed lines; from line 8 onward the file is a sequence of haplotype and/or genotype blocks.

Line	Contents
1	<pedigree_file> <output_file>
2	Three random number generator seeds, each less than 30000
3	Number of trait/affection columns between sex (column 5) and the marker mask
4	Number of replicates
5	Total number of marker loci, then the number of times the block pattern (lines 8+) is repeated to fill those loci
6	Inter-marker distance type: 1 recombination fractions, 2 Kosambi map distances (Morgans), 3 Haldane map distances (Morgans)
7	<count> <values…>: number of inter-marker distances supplied followed by the values themselves. See "Inter-marker distance expansion" below

Line 5 pattern repetition. totalLoci / iteration must be a whole number. If a block pattern covers 13 markers and 1300 loci are wanted, set line 5 to 1300 100; the blocks from line 8 onward are replayed 100 times.

Inter-marker distance expansion. The <count> field on line 7 admits three forms:

• count = 1: one value, used for every inter-marker interval.
• count = totalLoci − 1: every interval supplied explicitly.
• 1 < count < totalLoci − 1: the supplied values cycle as a repeating pattern across the totalLoci − 1 intervals.

Kosambi and Haldane map distances are converted internally to recombination fractions using $\theta = \tfrac{1}{2}(e^{4d}-1)/(e^{4d}+1)$ (Kosambi) and $\theta = \tfrac{1}{2}(1-e^{-2d})$ (Haldane), with $d$ in Morgans.

Haplotype block (line begins with 1)

1 <n_haplotypes> <n_loci> <n_repeats>
<freq_1> <freq_2> ... <freq_n_haplotypes>
<allele_1> <allele_2> ... <allele_n_loci>     ← haplotype 1 alleles
<allele_1> <allele_2> ... <allele_n_loci>     ← haplotype 2 alleles
...

Loci within a block are in linkage disequilibrium and share one set of haplotype frequencies; distinct blocks are in linkage equilibrium with one another.

Genotype block (line begins with 2)

2 <n_loci> <n_repeats>
<n_alleles_locus_1> <freq_a1> <freq_a2> ... <freq_a(n-1)>
<n_alleles_locus_2> <freq_a1> <freq_a2> ... <freq_a(n-1)>
...

Only n_alleles − 1 frequencies are listed per locus; the final frequency is inferred. Genotypes are simulated under Hardy-Weinberg equilibrium.

Haplotype and genotype blocks may alternate freely. The total number of marker loci across all blocks in one round of the pattern must equal totalLoci / iteration from line 5.

---

Worked examples

Each example below has a parameter file input<k>.dat and a pedigree file pedin<k>.pre in examples/. Examples 1 to 4 reuse one nuclear pedigree of six individuals (parents 1 and 2, children 3 to 6) with body mass index as the quantitative trait. Example 5 demonstrates a multi-family pedigree file.

Example 1: pure haplotypes

10 marker loci forming 8 haplotypes, all in linkage disequilibrium, 1000 replicates, recombination fractions supplied directly.

pedin1.pre pedfile1.pre  << name of pedigree file, name of output file
23221 1601 21001         << three random seeds
0                        << number of trait columns
1000                     << number of replicates
10 1                     << total markers, pattern repeats
1                        << recombination fractions on line 7
9 0.001 0.0 0.005 0.002 0.0 0.01 0.002 0.0 0.001
1 8 10 1                 << haplotype block: 8 haplotypes, 10 loci, repeat 1×
0.35 0.15 0.10 0.08 0.08 0.08 0.08 0.08          << haplotype frequencies
1 1 1 1 1 1 1 1 1 1                              << haplotype 1
2 2 2 2 1 1 1 1 1 1
1 1 2 2 2 2 2 1 1 2
1 2 1 2 1 2 1 2 1 2
1 1 1 1 2 2 2 2 2 1
2 2 2 1 1 1 1 2 2 2
1 1 2 2 1 1 2 2 1 1
2 2 2 1 1 2 2 2 2 1

Example 2: haplotypes with Kosambi map distances, per-individual mask, multiallelic markers

10 marker loci forming 10 haplotypes, 10000 replicates, BMI included as a quantitative trait. Markers 2 and 9 have more than two alleles. The pedigree file uses per-individual marker masks to blank specific markers for individuals 4 and 6:

1 1 0 0 1 22.3  0                          ← individual 1 deceased, all blanked
1 2 0 0 2 30.3  1                          ← all markers emitted
1 3 1 2 1 29.6  1
1 4 1 2 2 24.5  1 1 1 1 0 1 1 0 1 1        ← markers 5 and 8 blanked
1 5 1 2 1 35.7  1
1 6 1 2 2 23.5  1 1 0 1 1 1 1 1 0 0        ← markers 3, 9, 10 blanked

Parameter file input2.dat opens with:

pedin2.pre pedfile2.pre  << name of pedigree file, name of output file
21998 26002 11981        << three random seeds
1                        << one trait column (BMI)
10000                    << number of replicates
10 1                     << total markers, pattern repeats
2                        << Kosambi distances on line 7
9 0.001 0.02 0.005 0.02 0.001 0.001 0.002 0.01 0.001
1 10 10 1                << haplotype block: 10 haplotypes, 10 loci
0.30 0.25 0.10 0.10 0.05 0.04 0.04 0.04 0.04 0.04
1 3 1 1 1 1 1 1 4 1                        ← haplotype 1 (alleles 3, 4 multiallelic)
...

Example 3: pure genotypes under Hardy-Weinberg

20 marker loci in linkage equilibrium, mixing diallelic, triallelic, and hexallelic markers, 1000 replicates. Allele-frequency blocks are repeated to fill the 20 loci.

pedin3.pre pedfile3.pre  << name of pedigree file, name of output file
1344 2673 12228          << three random seeds
0                        << number of trait columns
1000                     << number of replicates
20 2                     << total markers, pattern repeats 2× to fill 20 loci
1                        << recombination fractions on line 7
10 0.01 0.05 0.05 0.09 0.01 0.01 0.02 0.05 0.01 0.05
2 3 2                    << genotype block: 3 loci, pattern repeated 2×
2 0.3                    << locus 1: 2 alleles, first allele freq 0.3
2 0.4                    << locus 2
2 0.2                    << locus 3
2 2 1                    << genotype block: 2 loci
3 0.7 0.1                << locus, 3 alleles
6 0.1 0.1 0.1 0.1 0.1    << locus, 6 alleles
2 1 2                    << genotype block: 1 locus, pattern repeated 2×
2 0.5

Example 4: mixture of haplotypes and genotypes

220 marker loci total, generated from a 22-marker block pattern repeated 10 times. The block contains one genotype block (6 markers) followed by two haplotype blocks (5 markers and 2 markers). This case exercises every code path: pattern repetition, alternating block types, masking via the pedigree file, and trait columns.

Example 5: multi-family pedigree file

examples/pedin5.pre interleaves two pedigrees (IDs 3 and 1) in one file. SimPed processes them independently; individual IDs need not be sequential within a family.

---

Output file

Output follows LINKAGE pedigree format. One line per individual per replicate:

<replicate_pedigree_id>  <ind_id>  <father>  <mother>  <sex>  [trait_columns…]  <a1_l1> <a2_l1>  <a1_l2> <a2_l2>  …

Replicates are concatenated; the pedigree-id column is re-numbered across replicates so the file remains a valid multi-pedigree LINKAGE file. The first two lines of pedfile1.pre from Example 1 (truncated to 20 markers) read:

1  1  0  0  1   0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1  2  0  0  2   2 2 2 2 2 2 1 1 1 1 2 1 2 1 2 2 2 2 1 2

For a post-MAKEPED file with first-offspring, paternal/maternal sibling, and proband columns, run the output through makeped from the LINKAGE package.

---

Algorithm

For each replicate the program executes the following.

1. Founder assignment. For each founder (both parental IDs 0) and each block of marker loci specified from line 8 onward, two haplotypes are drawn independently. In a haplotype block, a uniform random number is compared to the cumulative haplotype-frequency vector to select one of the supplied haplotype rows. In a genotype block, alleles at each locus are drawn independently from the supplied allele-frequency vector.
2. Meiotic drop-down. Within each pedigree, every individual whose parents are both filled is visited in turn. For each parental contribution (paternal and maternal), one of the parent's two haplotypes is selected uniformly at random at locus 1. At each subsequent locus j, a uniform draw is compared to the inter-marker recombination fraction $\theta_{j-1}$; with probability $\theta_{j-1}$ the haplotype of origin switches, otherwise it persists. The process repeats over all totalLoci loci. A while-loop over the family ensures that individuals are filled in topological order regardless of input ordering.
3. Masking and output. The pedigree file is read a second time; the marker mask column is consulted per individual to decide whether each locus is emitted as the simulated genotype or as 0 0. The pedigree identifier is re-numbered to remain unique across replicates.

The random number generator is a three-component Wichmann-Hill style generator seeded by line 2 of the parameter file. Identical seeds reproduce identical output.

---

Compile-time limits

Hard limits are set as #define constants in src/simped.c. Adjust and rebuild for larger inputs.

Constant	Default	Meaning
MAX_PED	100	maximum number of pedigrees
MAX_IND	3000	maximum number of individuals
MAX_LOCUS	10000	maximum number of loci
MAX_HAPLOTYPES	1000	maximum number of haplotypes
MAX_ALLELE	20	maximum number of alleles at a locus

---

Changes from the upstream source

The 2005 source did not compile on macOS or under recent gcc/clang because of a name collision and several implicit-int constructs that ceased to be valid C in C99.

• The global index was renamed to founder_marker_idx. The original name collides with the POSIX index() function declared in <strings.h>, which is transitively included on macOS and BSD platforms.
• Explicit return types were added to main() and writeResult().
• Forward declarations were added for all internal functions.
• Field-width caps were added to scanf formats and return values of fopen and fscanf are now checked.
• The haplotype-frequency cumulative-sum loop no longer reads possibilities[-1] on its first iteration.
• A small number of stack-allocated variable-length arrays were replaced with fixed-size buffers tied to the existing MAX_LOCUS and MAX_ALLELE constants.

---

License and contact

The upstream distribution did not include an explicit software license. The original program and documentation is © Suzanne M. Leal (2005). For redistribution terms or questions about the underlying algorithm, contact the original authors.