Intrinsic postzygotic hybrid incompatibilities are usually due to negative epistatic interactions between alleles from different parental genomes. While such incompatibilities are thought to be uncommon in speciation with gene flow, they may be important if such speciation results from a hybrid population. Here we aimed to test this idea in the endemic cichlid fishes of Lake Victoria. Hundreds of species have evolved within the lake in <15k years from hybrid progenitors. While the importance of prezygotic barriers to gene flow is well established in this system, the possible relevance of postzygotic genetic incompatibilities is unknown. We inferred the presence of negative epistatic interactions from systematic patterns of genotype ratio distortions in experimental crosses and wild samples. We then compared the positions of putative incompatibility loci to regions of high genetic differentiation between sympatric sister species as well as between members of clades that may have arisen in the early history of this radiation, and further determined if the loci showed fixed differences between the closest living relatives of the lineages ancestral to the hybrid progenitors. Overall, we find little evidence for a major role of intrinsic postzygotic incompatibilities in the Lake Victoria radiation. However, we find putative incompatibility loci significantly more often coinciding with islands of genetic differentiation between species that separated early in the radiation than between the younger sister species, consistent with the hypothesis that such variants segregated in the hybrid swarm and were sorted between species in the early speciation events.

F2 hybrid CROSS information (species used as female x species used as male)

PNPxPPP stands for the cross between Pundamilia sp. "nyererei-like" and Pundamilia sp. "pundamilia-like"
PPMxPRHZ stands for the cross between Pundamilia pundamilia and Pundamilia sp. "red-head"
PNPxNOM stands for the cross between Pundamilia sp. "nyererei-like" and Neochromis omnicaeruleus

Analysis steps with list of input and output files

CROSSES

1. filter cross VCF files using "Crosses_VCF_filtering.txt"

input: the unfiltered cross VCF files:

PNPxNOM_7more.vcf.gz -> originally from Feller et al. 2022, https://doi.org/10.5061/dryad.931zcrjm2
PNPxPPP_final.vcf.gz -> originally from Feller et al. 2020, https://doi.org/10.5061/dryad.5tb2rbp1n
PPMxPRHZ.vcf.gz -> originally from Feller et al. 2020, https://doi.org/10.5061/dryad.5tb2rbp1n

output: the quality filtered, phased&imputed, and subsetted VCF files (grandparental genotypes in a separate VCF file):

CROSS_f1_f2_homfixF1het_noscaffolds_phased.vcf.gz
CROSS_Pgenos.vcf.gz

info:

Crosses_VCF_filtering_individuals_to_exclude_overview.txt

2. get heterozygosity, allele frequencies, and genotype frequencies using "Crosses_Analyses.txt"

input: output from 1., the filtered VCF files:

CROSS_f1_f2_homfixF1het_noscaffolds_phased.vcf.gz

output: heterozygosity (het), allele frequency (frq), genotype frequency (hwe) tables:

CROSS.het
CROSS_phased.het
CROSS_fixed_phased_all.frq
CROSS_fixed_phased_all.hwe

3. use JoinMap to identify identical/highly similar loci

a. generate JoinMap input files using "Crosses_vcfToJoinMap.R"

input: output from 1., the filtered VCF files:

CROSS_f1_f2_homfixF1het_noscaffolds_phased.vcf.gz

output: JoinMap input files:

CROSS_JMinput.txt

b. in JoinMap, subset markers as described in "Crosses_VCF_filtering_JoinMap.txt"

input: output from 3a, JoinMap input files:

CROSS_JMinput.txt

output: list that indicates which loci to exclude:

CROSS_exclude975.txt

4. exclude identical/highly similar loci identified in JoinMap in "Crosses_getting_thinned_loci.R"

input: output from 3., the frq and hwe tables, and the grandparental genotype files from 1.:

CROSS_exclude975.txt
CROSS_fixed_phased_all.frq
CROSS_fixed_phased_all.hwe
CROSS_Pgenos.vcf.gz

output: subsetted allele frequency (frq), genotype frequency (hwe) tables, and grandparental genotype files:

CROSS_fixed_phased_all_sub.frq
CROSS_fixed_phased_all_sub.hwe
CROSS_Pgenos_sub.txt

5. Heteroyzgosity analysis using "Crosses_Analyses_Heterozygosity.R"

input: output from 2., the het tables:

CROSS.het
CROSS_phased.het

output: results tables

6. identify segregation distorted loci using "Crosses_Analyses_SegregationDistortion_CROSS.R"

input: output from 4., the subsetted frw and hwe, and the grandparental genotype files:

CROSS_fixed_phased_all_sub.frq
CROSS_fixed_phased_all_sub.hwe
CROSS_Pgenos_sub.txt

output: list of loci with any of the three patterns of segregation distortion per cross:

CROSS_ALLdeviations_sub.csv

WHOLE GENOMES (WG)

7. filtering of WG VCF files and LD analysis using "WholeGenomes_filtering_and_Analyses_LD.txt"

a. subset to 107 samples including ancestors and filter

input: whole genome VCF tables split up by chromosome, and list of samples:

allGenomes.chr${IDX}.SNPs.minDP6.minGQ20.max0.5N.3masks.vcf.gz -> generated by Meier et al. 2023, Science
WholeGenomes_LDsamples_107.txt

output: filtered whole genome VCF table with genotypes for 107 samples:

forLDscan.chrALL.vcf.gz

b. subset to 94 LV samples and filter some more including intra-LD pruning

input: filtered whole genome VCF tables split up by chromosome, and list of samples:

WholeGenomes_LDsamples_94.tx
forLDscan.chr${IDX}.vcf.gz

output: more stringently filtered whole genome VCF table with genotypes for 94 samples:

forLDscan.chrALL_pruned_mdp.vcf.gz

c. run LD analysis

input: output from 7b, the whole genome VCF table with genotypes for 94 samples:

forLDscan.chrALL_pruned_mdp.vcf.gz

output: results files from linkage disequilibrium (LD) analysis

LDscan.ld # not saved (huge file)
LDscan.ld_onlyinterchrom # not saved (huge file)
LDscan.ld_onlyinterchrom_sub[05/06]

d. check check genotypes of ancestors at high LD sites

input: output from 7a, the whole genome VCF table with genotypes for 107 samples, and the positions of loci in high intra-chromosomal LD:

forLDscan.chrALL.vcf.gz
LDpos_r06.txt # generated in FSTvsLD.R, see below (originally called LDpos.txt)
LDpos_r05.txt # generated in FSTvsLD.R, see below (originally called LDpos2.txt)

output: the whole genome VCF table with genotypes for 107 samples subsetted to high LD loci:

LDlocSubset107.vcf.gz #(r2>0.6)
LDlocSubset107_r05.vcf.gz #(r2>0.5)

8. filtering of WG VCF files and FST analysis using "WholeGenomes_filtering_and_Analyses_FST.txt"

input: whole genome VCF tables split up by chromosome, and list of samples:

allGenomes.chr${IDX}.SNPs.minDP6.minGQ20.max0.5N.3masks.vcf.gz -> generated by Meier et al. 2023, Science
WholeGenomes_FSTsamples.txt

output: more stringently filtered whole genome VCF table and results files of FST scans:

forFSTscans.chrALL_mdp.vcf.gz
FSTscanPairX.windowed.weir.fst

OVERLAPS

9. testing overlaps in segregation distortion vs LD using "DistortionsvsLD.R"

input: output from 6. and 7c., lists of loci with segregation distortion or in high intra-chromosomal LD:

CROSS_ALLdeviations_sub.csv
LDscan.ld_onlyinterchrom_sub06
LDscan.ld_onlyinterchrom_sub05

output:

Number of overlaps, p-values (not saved as files)
DistortionvsLD_Overlaps05_not_thinned.csv # for analysis 13.
DistortionvsLD_Overlaps06_not_thinned.csv # for analysis 13.

10. testing overlaps in segregation distortion vs using "FSTvsDistortions.R"

input: output from 6. and 8., lists of loci with segregation distortion and FST tables:

CROSS_ALLdeviations_sub.csv
chrInfo.txt -> from Feulner et al. 2018 G3, https://doi.org/10.5061/dryad.59q56g6
FSTscan_CROSS.windowed.weir.fst

output:

list of snps (not saved as file, only very few)
pvalues (not saved as file)

script also checks overlaps in segregation distorted snps among crosses

11. testing overlaps FST vs LD using "FSTvsLD.R" (same script with in/output files with 'ns' suffix for non-sister tests)

input: output from 7c. and 8., lists of loci in high intra-chromosomal LD and FST tables:

chrInfo.txt -> from Feulner et al. 2018 G3, https://doi.org/10.5061/dryad.59q56g6
LDscan.ld_onlyinterchrom_sub06
LDscan.ld_onlyinterchrom_sub05
FSTscanPairX.windowed.weir.fst

output:

LDpos_r05.txt # used in 7. to subset ancestor genotype files
LDpos_r06.txt # used in 7. to subset ancestor genotype files
LD_vs_FST_overlaps_06.txt
LD_vs_FST_overlaps_05.txt
FSTvsLD_Overlaps_uniqueSNPs_r06_no_thinning.csv # for analysis 13.
FSTvsLD_Overlaps_uniqueSNPs_r05_no_thinning.csv # for analysis 13.

script also includes Fisher's method to test if the combination of p-values depart from expectation

ANCESTOR GENOTYPES at high LD loci and GENES at overlapping loci

12. check which high LD loci are fixed in extant relatives of hybrid swarm ancestor using "LDloc_genotpyes.R"

input: output from 7d, the genotypes of extant relatives of hybrid swarm ancestors at high LD sites:

LDlocSubset107.vcf.gz
LDlocSubset107_r05.vcf.gz

output: list of loci in high LD and fixed between ancestors:

fixedLDloci_r05_V2.csv # 5 Congo vs 8 Nile
fixedLDloci_r06_V2.csv # 5 Congo vs 8 Nile

13. check which fixed high LD loci overlap with high FST or distorted regions using "LDlocfixed_vs_FSTandDistortions.R" (same script with in/output files with 'ns' suffix for non-sister tests)

input: output from 9., 11., 12.:

fixedLDloci_r05_V2.csv
fixedLDloci_r06_V2.csv
FSTvsLD_Overlaps_uniqueSNPs_r06_no_thinning.csv
FSTvsLD_Overlaps_uniqueSNPs_r05_no_thinning.csv
DistortionvsLD_Overlaps06_not_thinned.csv
DistortionvsLD_Overlaps05_not_thinned.csv
chrInfo.txt -> from Feulner et al. 2018 G3, https://doi.org/10.5061/dryad.59q56g6
FSTscanPair5.windowed.weir.fst

output: list of loci in high LD and in FST outlier window and fixed between ancestors:

LDandFSTandfixed_r05and06_NEW_non_thinned.csv

14. extract in which genes putative incompatibility loci are using "genes.R" (same script with in/output files with 'ns' suffix for non-sister tests)

input: output from 13., and the P. nyererei gff file

P_nyererei_v2.gff.gz -> from Feulner et al. 2018 G3, https://doi.org/10.5061/dryad.59q56g6
LDandFSTandfixed_r05and06_NEW_non_thinned.csv

output: list of genes:

genes_NEW_non_thinned.csv # added GO terms manually (from UniProt via Ensembl), see Table S8

PLOTTING and OTHER

FSTscans_similarityTests.R (also tests correlations with recombination landscape)
Plots.R
Plots_FST.R

input:

FSTscanPairX.windowed.weir.fst, P_nyererei_v2.RecRates.txt -> from Feulner et al. 2018 G3, https://doi.org/10.5061/dryad.59q56g6

output:

FSTscan_correlations.txt

Note: 11., 13., 14. also run for non sister pairs in a separate script (suffix _ns in files)

Data from: Testing for a role of postzygotic incompatibilities in rapidly speciated Lake Victoria cichlids

Data files

Abstract

README: Testing for a role of postzygotic incompatibilities in rapidly speciated Lake Victoria cichlids

F2 hybrid CROSS information (species used as female x species used as male)

Analysis steps with list of input and output files

CROSSES

1. filter cross VCF files using "Crosses_VCF_filtering.txt"

2. get heterozygosity, allele frequencies, and genotype frequencies using "Crosses_Analyses.txt"

3. use JoinMap to identify identical/highly similar loci

4. exclude identical/highly similar loci identified in JoinMap in "Crosses_getting_thinned_loci.R"

5. Heteroyzgosity analysis using "Crosses_Analyses_Heterozygosity.R"

6. identify segregation distorted loci using "Crosses_Analyses_SegregationDistortion_CROSS.R"

WHOLE GENOMES (WG)

7. filtering of WG VCF files and LD analysis using "WholeGenomes_filtering_and_Analyses_LD.txt"

8. filtering of WG VCF files and FST analysis using "WholeGenomes_filtering_and_Analyses_FST.txt"

OVERLAPS

9. testing overlaps in segregation distortion vs LD using "DistortionsvsLD.R"

10. testing overlaps in segregation distortion vs using "FSTvsDistortions.R"

11. testing overlaps FST vs LD using "FSTvsLD.R" (same script with in/output files with 'ns' suffix for non-sister tests)

ANCESTOR GENOTYPES at high LD loci and GENES at overlapping loci

12. check which high LD loci are fixed in extant relatives of hybrid swarm ancestor using "LDloc_genotpyes.R"

13. check which fixed high LD loci overlap with high FST or distorted regions using "LDlocfixed_vs_FSTandDistortions.R" (same script with in/output files with 'ns' suffix for non-sister tests)

14. extract in which genes putative incompatibility loci are using "genes.R" (same script with in/output files with 'ns' suffix for non-sister tests)

PLOTTING and OTHER

Works referencing this dataset