void GenotypeList::Print(GenotypeList * list, Pedigree & ped, Family * family, int marker)
{
MarkerInfo * info = ped.GetMarkerInfo(marker);
for (int i = 0; i < family->count; i++)
{
printf("%s - ", (const char *) ped[family->path[i]].pid);
for (int j = 0; j < list[i].allele1.Length(); j++)
{
if (list[i].allele1[j] == -1)
printf("*/");
else
printf("%s/", (const char *) info->GetAlleleLabel(list[i].allele1[j]));
if (list[i].allele2[j] == -1)
printf("* ");
else
printf("%s ", (const char *) info->GetAlleleLabel(list[i].allele2[j]));
}
printf("\n");
}
printf("\n");
}
void DosageCalculator::OutputBasicMarkerInfo(FILE * output)
{
fprintf(output, "SNP\tAl1\tAl2\tFreq\tRsq_hat\n");
double scale = 1.0 / (genotypes + 1e-30);
for (int marker = 0; marker < markers; marker++)
{
double sum = 0.0, sumsq = 0.0;
for (int sample = 0; sample < genotypes; sample++)
{
double dose = GetDosage(sample, marker);
sum += dose;
sumsq += dose * dose;
}
sum *= scale;
sumsq *= scale;
double freq = sum * 0.50;
double var1 = 2 * freq * (1.0 - freq);
double var2 = max(sumsq - sum * sum, 0.0);
MarkerInfo * info = Pedigree::GetMarkerInfo(marker);
fprintf(output, "%s\t%s\t%s\t%.4f\t%.4f\n",
(const char *) info->name,
(const char *) info->GetAlleleLabel(1),
info->CountAlleles() > 1 ? (const char *) info->GetAlleleLabel(2) : "-",
freq > 0.50 ? 1.0 - freq : freq, var2 > var1 ? 1.0 : var2 / (var1 + 1e-30));
}
}
void DosageCalculator::OutputMarkerInfo(FILE * output)
{
if (stored == 0) return;
if (!storeDistribution)
{
OutputBasicMarkerInfo(output);
return;
}
fprintf(output, "SNP\tAl1\tAl2\tFreq1\tMAF\tQuality\tRsq\n");
double scale_sg = 1.0 / (samples * genotypes + 1e-30);
double scale_g = 1.0 / (genotypes + 1e-30);
double scale_ss = 1.0 / (samples * samples + 1e-30);
for (int marker = 0; marker < markers; marker++)
{
double p0 = 0.0, p1 = 0.0;
double qc = 0.0, sumsq = 0.0;
for (int sample = 0; sample < genotypes; sample++)
{
unsigned int n0, n1, n2;
GetCounts(sample, marker, n0, n1, n2);
p0 += n0; p1 += n1;
qc += (n0 > n1 && n0 > n2) ? n0 : (n1 > n2) ? n1 : n2;
sumsq += square(n0 + n1 * 0.5) * scale_ss;
}
p0 *= scale_sg; p1 *= scale_sg;
qc *= scale_sg; sumsq *= scale_g;
double freq = p0 + p1 * 0.50;
double var1 = max(p0 + p1 * 0.25 - square(freq), 0);
double var2 = max(sumsq - square(freq), 0);
MarkerInfo * info = Pedigree::GetMarkerInfo(marker);
fprintf(output, "%s\t%s\t%s\t%.4f\t%.4f\t%.4f\t%.4f\n",
(const char *) info->name,
(const char *) info->GetAlleleLabel(1),
info->CountAlleles() > 1 ? (const char *) info->GetAlleleLabel(2) : "-",
freq, freq > 0.50 ? 1.0 - freq : freq, qc, var2 / (var1 + 1e-30));
}
}
bool Pedigree::SexLinkedCheck()
{
bool fail = false;
// Keep track of what families fail the basic inheritance check,
// so that we can run later run genotype elimination check on the remainder
IntArray failedFamilies(familyCount);
// For each marker ...
for (int m = 0; m < markerCount; m++)
{
MarkerInfo * info = GetMarkerInfo(m);
failedFamilies.Zero();
// Check for homozygous males
for (int f = 0; f < familyCount; f++)
for (int i = families[f]->first; i <= families[f]->last; i++)
if (persons[i]->sex == SEX_MALE && persons[i]->markers[m].isKnown() &&
!persons[i]->markers[m].isHomozygous())
{
printf("%s - Fam %s: Male %s has two X alleles [%s/%s]\n",
(const char *) markerNames[m],
(const char *) persons[i]->famid, (const char *) persons[i]->pid,
(const char *) info->GetAlleleLabel(persons[i]->markers[m].one),
(const char *) info->GetAlleleLabel(persons[i]->markers[m].two));
// Wipe this genotype so we don't get cascading errors below
persons[i]->markers[m][0] = persons[i]->markers[m][1] = 0;
fail = true;
failedFamilies[f] = true;
}
// Check full sibships for errors
// TODO -- We could do better by grouping male half-sibs
for (int f = 0; f < familyCount; f++)
for (int i = families[f]->first; i <= families[f]->last; i++)
if (!persons[i]->isFounder() && persons[i]->sibs[0] == persons[i])
{
// This loop runs once per sibship
Alleles fat = persons[i]->father->markers[m];
Alleles mot = persons[i]->mother->markers[m];
bool fgeno = fat.isKnown();
bool mgeno = mot.isKnown();
Alleles inferred_mother = mot;
Alleles first_sister;
Alleles inferred_father;
bool mother_ok = true;
int sisters = 0;
for (int j = 0; j < persons[i]->sibCount; j++)
if (persons[i]->sibs[j]->isGenotyped(m))
{
Alleles geno = persons[i]->sibs[j]->markers[m];
bool fat1 = fat.hasAllele(geno.one);
bool fat2 = fat.hasAllele(geno.two);
bool mot1 = mot.hasAllele(geno.one);
bool mot2 = mot.hasAllele(geno.two);
int sex = persons[i]->sibs[j]->sex;
if (sex == SEX_MALE)
{
if (mgeno && !mot1)
{
printf("%s - Fam %s: Child %s [%s/Y] has mother [%s/%s]\n",
(const char *) markerNames[m],
(const char *) persons[i]->famid,
(const char *) persons[i]->sibs[j]->pid,
(const char *) info->GetAlleleLabel(geno.one),
(const char *) info->GetAlleleLabel(mot.one),
(const char *) info->GetAlleleLabel(mot.two));
fail = true;
failedFamilies[f] = true;
}
else
mother_ok &= inferred_mother.AddAllele(geno.one);
}
if (sex == SEX_FEMALE)
{
if ((fgeno && mgeno && !((fat1 && mot2) || (fat2 && mot1))) ||
(fgeno && !(fat1 || fat2)) || (mgeno && !(mot1 || mot2)))
{
printf("%s - Fam %s: Child %s [%s/%s] has ",
(const char *) markerNames[m],
(const char *) persons[i]->famid,
(const char *) persons[i]->sibs[j]->pid,
(const char *) info->GetAlleleLabel(geno.one),
(const char *) info->GetAlleleLabel(geno.two));
if (!fgeno)
printf("mother [%s/%s]\n",
(const char *) info->GetAlleleLabel(mot.one),
(const char *) info->GetAlleleLabel(mot.two));
else if (!mgeno)
printf("father [%s/Y]\n",
(const char *) info->GetAlleleLabel(fat.one));
else
printf("parents [%s/Y]*[%s/%s]\n",
(const char *) info->GetAlleleLabel(fat.one),
(const char *) info->GetAlleleLabel(mot.one),
(const char *) info->GetAlleleLabel(mot.two));
fail = true;
failedFamilies[f] = true;
}
else
{
if (!sisters++)
inferred_father = first_sister = geno;
else if (first_sister != geno)
{
inferred_father.Intersect(geno);
mother_ok &= inferred_mother.AddAllele(
geno.otherAllele(inferred_father.one));
mother_ok &= inferred_mother.AddAllele(
first_sister.otherAllele(inferred_father.one));
}
if (!fgeno && (mot1 ^ mot2))
inferred_father.Intersect(mot1 ? geno.two : geno.one);
if (!mgeno && (fat1 ^ fat2))
mother_ok &= inferred_mother.AddAllele(fat1 ? geno.two : geno.one);
}
}
}
if (!mother_ok || (sisters && !inferred_father.isKnown()))
{
printf("%s - Fam %s: ",
(const char *) markerNames[m],
(const char *) persons[i]->famid);
if (fgeno)
printf("Father %s [%s/Y] has children [",
(const char *) persons[i]->father->pid,
(const char *) info->GetAlleleLabel(fat.one));
else if (mgeno)
printf("Mother %s [%s/%s] has children [",
(const char *) persons[i]->mother->pid,
(const char *) info->GetAlleleLabel(mot.one),
(const char *) info->GetAlleleLabel(mot.two));
else
printf("Couple %s * %s has children [",
(const char *) persons[i]->mother->pid,
(const char *) persons[i]->father->pid);
for (int j = 0; j < persons[i]->sibCount; j++)
printf(
persons[i]->sibs[j]->sex == SEX_MALE ? "%s%s/Y" : "%s%s/%s",
j == 0 ? "" : " ",
(const char *) info->GetAlleleLabel(persons[i]->sibs[j]->markers[m].one),
(const char *) info->GetAlleleLabel(persons[i]->sibs[j]->markers[m].two));
printf("]\n");
fail = true;
failedFamilies[f] = true;
}
}
for (int f = 0; f < familyCount; f++)
if (!failedFamilies[f] &&
(families[f]->count > families[f]->founders + 1) &&
!families[f]->isNuclear())
fail |= !GenotypeList::EliminateGenotypes(*this, families[f], m);
}
if (fail)
printf("\nMendelian inheritance errors detected\n");
return fail;
}
bool Pedigree::AutosomalCheck()
{
// Arrays indicating which alleles and homozygotes occur
IntArray haplos, genos, counts, failedFamilies;
bool fail = false;
// For each marker ...
for (int m = 0; m < markerCount; m++)
{
MarkerInfo * info = GetMarkerInfo(m);
// Summary for marker
int alleleCount = CountAlleles(m);
int genoCount = alleleCount * (alleleCount + 1) / 2;
// Initialize arrays
haplos.Dimension(alleleCount + 1);
haplos.Set(-1);
genos.Dimension(genoCount + 1);
genos.Set(-1);
failedFamilies.Dimension(familyCount);
failedFamilies.Zero();
counts.Dimension(alleleCount + 1);
for (int f = 0; f < familyCount; f++)
for (int i = families[f]->first; i <= families[f]->last; i++)
if (!persons[i]->isFounder() && persons[i]->sibs[0] == persons[i])
{
// This loop runs once per sibship
Alleles fat = persons[i]->father->markers[m];
Alleles mot = persons[i]->mother->markers[m];
bool fgeno = fat.isKnown();
bool mgeno = mot.isKnown();
// Number of alleles, homozygotes and genotypes in this sibship
int haplo = 0, h**o = 0, diplo = 0;
// No. of different genotypes per allele
counts.Zero();
// In general, there should be no more than 3 genotypes per allele
bool too_many_genos = false;
for (int j = 0; j < persons[i]->sibCount; j++)
if (persons[i]->sibs[j]->isGenotyped(m))
{
Alleles geno = persons[i]->sibs[j]->markers[m];
int fat1 = fat.hasAllele(geno.one);
int fat2 = fat.hasAllele(geno.two);
int mot1 = mot.hasAllele(geno.one);
int mot2 = mot.hasAllele(geno.two);
if ((fgeno && mgeno && !((fat1 && mot2) || (fat2 && mot1))) ||
(fgeno && !(fat1 || fat2)) || (mgeno && !(mot1 || mot2)))
{
printf("%s - Fam %s: Child %s [%s/%s] has ",
(const char *) markerNames[m],
(const char *) persons[i]->sibs[j]->famid,
(const char *) persons[i]->sibs[j]->pid,
(const char *) info->GetAlleleLabel(geno.one),
(const char *) info->GetAlleleLabel(geno.two));
if (!fgeno || !mgeno)
printf("%s [%s/%s]\n",
fgeno ? "father" : "mother",
(const char *) info->GetAlleleLabel(fgeno ? fat.one : mot.one),
(const char *) info->GetAlleleLabel(fgeno ? fat.two : mot.two));
else
printf("parents [%s/%s]*[%s/%s]\n",
(const char *) info->GetAlleleLabel(fat.one),
(const char *) info->GetAlleleLabel(fat.two),
(const char *) info->GetAlleleLabel(mot.one),
(const char *) info->GetAlleleLabel(mot.two));
fail = true;
failedFamilies[f] = true;
}
else
{
if (haplos[geno.one] != i)
{
haplo++;
haplos[geno.one] = i;
};
if (haplos[geno.two] != i)
{
haplo++;
haplos[geno.two] = i;
};
int index = geno.SequenceCoded();
if (genos[index] != i)
{
genos[index] = i;
diplo++;
counts[geno.one]++;
if (geno.isHomozygous())
h**o++;
else
counts[geno.two]++;
if (counts[geno.one] > 2) too_many_genos = true;
if (counts[geno.two] > 2) too_many_genos = true;
}
}
}
if (fgeno)
{
if (haplos[fat.one] != i)
{
haplo++;
haplos[fat.one] = i;
}
if (haplos[fat.two] != i)
{
haplo++;
haplos[fat.two] = i;
}
h**o += fat.isHomozygous();
}
if (mgeno)
{
if (haplos[mot.one] != i)
{
haplo++;
haplos[mot.one] = i;
}
if (haplos[mot.two] != i)
{
haplo++;
haplos[mot.two] = i;
}
h**o += mot.isHomozygous();
}
if (diplo > 4 || haplo + h**o > 4 || (haplo == 4 && too_many_genos))
{
printf("%s - Fam %s: ",
(const char *) markerNames[m],
(const char *) persons[i]->famid);
if (persons[i]->father->markers[m].isKnown())
printf("Father %s [%s/%s] has children [",
(const char *) persons[i]->father->pid,
(const char *) info->GetAlleleLabel(fat.one),
(const char *) info->GetAlleleLabel(fat.two));
else if (persons[i]->mother->markers[m].isKnown())
printf("Mother %s [%s/%s] has children [",
(const char *) persons[i]->mother->pid,
(const char *) info->GetAlleleLabel(mot.one),
(const char *) info->GetAlleleLabel(mot.two));
else
printf("Couple %s * %s has children [",
(const char *) persons[i]->mother->pid,
(const char *) persons[i]->father->pid);
for (int j = 0; j < persons[i]->sibCount; j++)
printf("%s%s/%s", j == 0 ? "" : " ",
(const char *) info->GetAlleleLabel(persons[i]->sibs[j]->markers[m].one),
(const char *) info->GetAlleleLabel(persons[i]->sibs[j]->markers[m].two));
printf("]\n");
fail = true;
failedFamilies[f] = true;
}
}
for (int f = 0; f < familyCount; f++)
if (!failedFamilies[f] &&
(families[f]->count > families[f]->founders + 1) &&
!families[f]->isNuclear())
fail |= !GenotypeList::EliminateGenotypes(*this, families[f], m);
}
if (fail)
printf("\nMendelian inheritance errors detected\n");
return fail;
}
