The literature

This page exists so that nobody has to take the article on trust.  Every source it rests on is listed here with what it actually measured, who wrote it, and which way it cuts.

It is deliberately not a reading list of things that agree with this site.  The two sources that carry most of the article's weight against the creationist reading, Wang 2007 and Lavialle 2013, are the same two that supply the creationist side's best material.  The creationist papers are tagged by what their data show, not by what their authors believe, and Tomkins's 2015 correction of his own method is tagged supports because it is honest work that happens to remove three of his own numbers.

How to read an entry

Measurement or model.  A count of collisions in a sequencing run is an observation.  An effective site pool derived from that count is a model, and it inherits every assumption in it.  Both are useful and they are not the same kind of thing.

Direction.  Assigned by what the source shows, not by what its author believes.

Verified.  Whether this project read the text directly (full), only its abstract (abstract), or only through another source that quotes it (via).  Nothing is quoted in the main article unless it is marked full or abstract.

Field definitions

Coverage and its limits, stated plainly.  This catalogue covers 24 sources:  the human-chimpanzee ERV census, the integration-preference literature, the three ERV function results, the four creationist pseudogene and method papers, and the olfactory-receptor dataset that shows both patterns at once.  Four items could not be obtained and are listed at the bottom;  two of them, Lachapelle & Drouin 2011 and Ma 2021, carry results the article's verdict depends on and should be fetched.

How many, and where

Initial sequence of the chimpanzee genome and comparison with the human genome

The Chimpanzee Sequencing and Analysis Consortium.  Nature 437(7055):69–87, 2005.  https://doi.org/10.1038/nature04072

camp: mainstream topic: erv-count method: measurement direction: undercuts verified: full

Key figures: Table 2, lineage-specific transposable-element activity in the aligned genomes — ERV class 1, 234 chimpanzee-specific against 5 human-specific;  ERV class 2, 45 against 77.  Sum 361.  HERV-K active in both lineages, at least 73 human-specific and 45 chimpanzee-specific insertions.

The reference count for the whole argument, and a lopsided one:  a few hundred unshared insertions against a background of order 10⁵ shared loci.  Two cautions the article prints.  Table 2 covers the aligned regions only, so it is not a whole-genome census.  And the widely repeated total of 391 is an arithmetic slip;  the four cells sum to 361.  This paper also supplies the creationist side's one genuinely awkward datum, PtERV1's absence from human and orangutan.

Long-term reinfection of the human genome by endogenous retroviruses

Belshaw, Robert, Vini Pereira, Aris Katzourakis, Gillian Talbot, Jan Pačes, Austin Burt & Michael Tristem.  PNAS 101(14):4894–4899, 2004.  https://doi.org/10.1073/pnas.0307800101

camp: mainstream topic: erv-count method: measurement direction: complicates verified: full

Key figures: ~98,000 HERV elements and fragments, 5 to 8% of the genome;  env dN/dS on internal branches 0.26 against 0.79 on terminal branches in HERV-K (p. 4897), and significantly below 1 on internal branches in seven of eight other families (Table 2, p. 4898:  HERV-E 0.22, HERV-H 0.24, HERV-T 0.12, HERV-R 0.45, HML3 0.73, HML5 0.40, HML6 0.35);  >12 HERV-K elements integrated since the human-chimpanzee split and at least two polymorphic in living humans;  solo LTRs 10 to 100 times more numerous than intact proviruses.

Cuts both ways and the creationist side has to be careful which half it quotes.  It bears on the ERV-first hypothesis less than this entry once said.  The env constraint sits on the internal branches of the copy tree and relaxes on the terminal ones, the signature of a lineage that kept infecting.  But Liu & Soper give env host jobs and propose designed transmission between cells with decay since the Fall, which fits the same shape;  and the earlier wording here, "the one gene with no possible host function", was wrong, since syncytin-1 and -2 are env genes.  Corrected in the article on 2026-10-27.  It supports a compressed timescale on a different point:  the family stayed infectious throughout and is still polymorphic, so insertion is not a frozen ancient event.  The 98,000 figure is Belshaw's citation to the HERVd database, not his own count.

Identification, characterization, and comparative genomic distribution of the HERV-K (HML-2) group

Subramanian, Ravi P., Julia H. Wildschutte, Crystal Russo & John M. Coffin.  Retrovirology 8:90, 2011.  https://doi.org/10.1186/1742-4690-8-90

camp: mainstream topic: erv-count method: measurement direction: complicates verified: full

Key figures: 91 full-length and near-full-length HML-2 proviruses plus 944 solo LTRs in GRCh37;  ~60% of elements in or within 30 kb of genes;  ~20% inside introns, ~80% of those antisense;  chromosome 17 gene-rich and devoid of HML-2.

The field's most careful census of the youngest HERV family, from John Coffin's lab.  Complicates both sides.  A tally made for this article from the "oldest common ancestor" column of Tables 1 and 2 gives 30 human-specific, 12 human-chimpanzee, 6 gorilla, 25 orangutan, 3 gibbon and 10 rhesus, out of 86 rows that parsed.  So about a third of HML-2 proviruses are human-specific, which is against the popular version of the shared-ERV argument, and the other two-thirds sort into a nested series by node, which is against the creationist one.  Any article quoting a "HERV-K count" has to say which of 91, 944 or 1,035 it means.  CC BY.

Lineage-specific expansions of retroviral insertions within the genomes of African great apes but not humans and orangutans

Yohn, Christopher T., Zhaoshi Jiang, Sean D. McGrath et al. and Evan E. Eichler.  PLoS Biology 3(4):e110, 2005.  https://doi.org/10.1371/journal.pbio.0030110

camp: mainstream topic: erv-count method: measurement direction: complicates verified: full

Key figures: 287 PtERV1 integration sites mapped across chimpanzee, gorilla, baboon and macaque;  95.8% at non-orthologous positions;  two of three apparent chimpanzee-macaque overlaps refined to single base pairs and found non-orthologous;  4 to 5 bp AT-rich target-site duplications in both.

The paper most often misread by both sides.  Its "six sites deleted in the human lineage … unlikely" sentence is Eichler's argument that the apparent sharing is not real, and he tests it in the next paragraph.  Quoted alone it reverses the paper's meaning.  What it genuinely gives the creationist case is that apparent sharing at map resolution can dissolve under base-pair refinement, and that independent clustering from integration bias is real.  What it costs that case is that the same test, applied to human-chimpanzee HERV-K, does not dissolve the sharing.  Its incongruent retroviral phylogeny is a real complication for the mainstream and Eichler names the alternatives.  CC BY.

Rate of recombinational deletion among human endogenous retroviruses

Belshaw, Robert, Jonathon Watson, Aris Katzourakis, Adam Howe, Jonathan Woolven-Allen, Austin Burt & Michael Tristem.  Journal of Virology 81(17):9437–9442, 2007.  https://doi.org/10.1128/JVI.02216-06

camp: mainstream topic: erv-count method: measurement direction: supports verified: abstract

Key figure: deletion among recently integrated HERV-K loci "almost 200-fold higher" than among loci predating the human-chimpanzee divergence, 8 × 10⁻⁵ against 4 × 10⁻⁷ events per locus per generation.

Cited for one point:  ERVs are removed cleanly at a measurable rate, so the absence of an ERV where the tree predicts one is weak evidence.  The fuller argument belongs to Shared Genetic Errors and is not repeated in the article.

Insertion-site preference

Retroviral DNA integration: ASLV, HIV, and MLV show distinct target site preferences

Mitchell, Rick S., Beth F. Beitzel, Astrid R. W. Schröder et al. and Frederic D. Bushman.  PLoS Biology 2(8):e234, 2004.  https://doi.org/10.1371/journal.pbio.0020234

camp: mainstream topic: insertion-preference method: measurement direction: supports verified: full

Key figures: HIV 75 to 80% of integrations in transcription units, MLV 61%, ASLV 57%, against 45% of the genome;  MLV strongly biased toward transcription start sites (p = 1.4 × 10⁻¹⁴), ASLV not (p > 0.05).

Supports the creationist premise and, quantified, defeats the conclusion drawn from it.  Integration is not random, so the "one in three billion" argument should never be used.  But a 75%-against-45% enrichment is a bias of 1.7-fold, and even MLV's start-site preference concentrates integration into 10⁷ to 10⁸ base pairs, not 10⁵.  CC BY.

HIV integration site selection: analysis by massively parallel pyrosequencing

Wang, Gary P., Angela Ciuffi, Jeremy Leipzig, Charles C. Berry & Frederic D. Bushman.  Genome Research 17(8):1186–1194, 2007.  https://doi.org/10.1101/gr.6286907

camp: mainstream topic: insertion-preference method: measurement direction: supports verified: full

Key figures: 40,569 unique HIV integration sites;  41 sites hosting two independent integrations at exactly the same base pair, counting opposite-orientation pairs only.

The strongest empirical card the creationist side has on this question, and the input to the article's own computation.  One collision per 990 integrations is about 150 times the random rate over a 3.1 Gb genome, so the popular claim that two viruses could never hit the same base pair is refuted by measurement.  The authors attribute the collisions to local sequence and chromosomal environment, not coincidence.  Its limit, printed in the article beside the result:  one virus, one immortalised T-cell line, not a germline-transmitted retrovirus in an ape.  Fetched with bin/pmcget.py after plain curl returned HTML from both the journal and PMC.

Function, and what it does and does not show

Paleovirology of "syncytins", retroviral env genes exapted for a role in placentation

Lavialle, Christian, Guillaume Cornelis, Anne Dupressoir, Cécile Esnault, Odile Heidmann, Cécile Vernochet & Thierry Heidmann.  Philosophical Transactions of the Royal Society B 368:20120507, 2013.  https://doi.org/10.1098/rstb.2012.0507

camp: mainstream topic: erv-function method: review direction: complicates verified: full

Key figures: six independently captured syncytins — syncytin-1 and -2 in primates, -A and -B in mice, -Ory1 in rabbits, -Car1 in carnivores;  mouse knockouts embryonic-lethal;  syncytin-Car1 in all 26 carnivore species tested and older than 80 Myr.

Supports the creationist function claim completely and destroys the inference usually drawn from it.  This is the article's control experiment:  Heidmann's group knows the primate and mouse syncytins were captured independently because they are not syntenic, which is exactly the test the shared-ERV argument applies.  Independent capture produced six different loci.  Peter Borger's "six times convergent evolution" objection therefore misfires as an answer to position.  Note also that syncytin-1 sits inside a full ERV-W provirus with LTRs, gag and pol remnants and a target-site duplication:  a co-opted insertion is still an insertion.

Regulatory evolution of innate immunity through co-option of endogenous retroviruses

Chuong, Edward B., Nels C. Elde & Cédric Feschotte.  Science 351(6277):1083–1087, 2016.  https://doi.org/10.1126/science.aad5497

camp: mainstream topic: erv-function method: measurement direction: complicates verified: full

Key figures: CRISPR deletion of the MER41 element 220 bp upstream of AIM2 abolished interferon-gamma induction of AIM2 protein entirely;  MER41-like gammaretroviruses colonised several mammalian lineages independently ~50 to 75 Myr ago;  a separate muroid ERV, RLTR30B, does the same job in mouse.

Same structure as the syncytin result and the same double edge.  ERVs are load-bearing regulatory DNA, proved by deletion.  And the paper establishes the independence of the enhancer families the same way:  different elements in different places.  The phrase to avoid is "ERVs are functional, therefore not viral" — the mechanism Chuong describes is a retroviral LTR carrying its own STAT1 motifs into new sites.  The extraction on disk is the author manuscript, so quotations are located by section rather than folio.

The vitamin C gene

The whole structure of the human nonfunctional L-gulono-γ-lactone oxidase gene

Inai, Yoko, Yoshitaka Ohta & Morimitsu Nishikimi.  Journal of Nutritional Science and Vitaminology 49(5):315–319, 2003.  https://doi.org/10.3177/jnsv.49.315

camp: mainstream topic: gulo method: measurement direction: supports verified: full

Key figures: five of the rat gene's twelve exons remain in human;  two single-base deletions, one three-base deletion, one insertion, two stop codons and two broken splice sites;  human and guinea pig share the identical substitution at 47 of the 129 positions where human differs from rat, 36%, p = 1.84 × 10⁻¹².

The single strongest published datum for the creationist side in this whole catalogue, and it comes from a mainstream lab with no stake in the debate.  Two lineages that lost GULO independently converged on the identical change at more than a third of the substituted positions, and Nishikimi's own conclusion is "many hot spots."  Its limits are printed in the same breath:  substitutions have four states per site, insertions have of order 10⁹ positions, and the rate is inflated by transition bias and CpG hypermutability.  Tomkins cites this paper and never mentions this result.

The human GULO pseudogene — evidence for evolutionary discontinuity and genetic entropy

Tomkins, Jeffrey P.  Answers Research Journal 7:91–101, 2014.

camp: creationist topic: gulo method: measurement direction: undercuts verified: full

Key figures: the 28,800-base human GULO region 84% identical to chimpanzee and 87% to gorilla;  the 13,000 bases upstream 68% and 73%;  exon-by-exon trees in which only exon 5 recovers the standard phylogeny;  differing transposable-element content in the four degenerate zones.

The strongest published creationist treatment of GULO, and three of its four claims do not survive.  The 84% and 68% figures were computed by the BLASTN method Tomkins himself documented as faulty the next year, and have never been recomputed.  The exon trees are built from 96- to 164-base sequences with no confidence measure, where incongruence is the expected result under any model.  The fourth claim, that the exon losses were independent in each taxon, may hold and is recorded as untested because its source could not be obtained.  His own Table 2 is his strongest evidence against himself:  every retained exon is 97.6 to 99.4% identical between human and chimpanzee and he offers no model that predicts that.  The ARJ PDF has a broken font encoding that destroys every digit;  the numbers were read from the journal HTML.

Documented anomaly in recent versions of the BLASTN algorithm

Tomkins, Jeffrey P.  Answers Research Journal 8:379–390, 2015.

camp: creationist topic: gulo method: measurement direction: supports verified: full

Key figures: only BLASTN 2.2.25+ returned hits for nearly all queries;  every later version omitted significant query sequence, worsening with each release;  the whole-genome similarity figure revised from 70% to 88% by nucmer, and 73% by LASTZ.

Tagged supports because it is honest work by a creationist that found and published an error in his own results.  Its effect on this article is destructive to his side:  it kills the 70% figure by its author's hand and, by extension, every number in the 2014 GULO paper computed the same way.  The 15-point disagreement between nucmer and LASTZ is itself the point, and belongs to the Human and Chimpanzee DNA Similarity article rather than this one.

The genetics of vitamin C loss in vertebrates

Drouin, Guy, Jean-Rémi Godin & Benoît Pagé.  Current Genomics 12(5):371–378, 2011.  https://doi.org/10.2174/138920211796429736

camp: mainstream topic: gulo method: review direction: complicates verified: full

Key claims: anthropoid primates and guinea pigs lost different exons;  bats show that inactive genes can be reactivated;  in passerines vitamin C synthesis was reacquired three times and lost twice.

Considered and used only by reference.  This material already appears in Shared Genetic Errors and the article links there rather than repeating it.  CC BY-NC.

The beta-globin pseudogene

Evolutionary constraints in the β-globin cluster: the signature of purifying selection at HBD

Moleirinho, Ana, Susana Seixas, Alexandra M. Lopes, Celeste Bento, Maria J. Prata & António Amorim.  Genome Biology and Evolution 5(3):559–571, 2013.  https://doi.org/10.1093/gbe/evt029

camp: mainstream topic: hbbp1 method: measurement direction: complicates verified: full

Key figures: HBD and HBBP1 have the lowest nucleotide diversity and most negative Tajima's D in the cluster, from 1,092 individuals plus 10 chimpanzees;  Table 3 human-chimpanzee divergence in HBBP1 exons 1.45%, against HBB 0.26% and HBD 0.80%.

The paper both sides should be reading.  It genuinely establishes that HBBP1 is under purifying selection and is not free-running junk, which is a real win for the function side.  It also contains two facts that damage the specific creationist argument built on it:  it says HBD and HBBP1 are "either marginally or not transcribed at all," which contradicts Tomkins's central mechanism and which he does not report;  and its Table 3 shows HBBP1's between-species divergence is higher than the working genes beside it, so the constraint signal is in within-species diversity, not divergence.  CC BY-NC.

Comparative analysis of three-dimensional chromosomal architecture identifies a novel fetal hemoglobin regulatory element

Huang, Peng, Cheryl A. Keller, Belinda Giardine et al. and Gerd A. Blobel.  Genes & Development 31(16):1704–1713, 2017.  https://doi.org/10.1101/gad.303461.117

camp: mainstream topic: hbbp1 method: measurement direction: complicates verified: full

Key figures: deleting the HBBP1 region in adult erythroid cells rewires contacts toward the fetal pattern and strongly raises γ-globin transcription;  deleting only the HBBP1 transcription start site abolishes the transcript and leaves globin transcription at control levels.

The decisive experiment on HBBP1, and neither side of the creation debate appears to have used it.  It splits the question exactly in half.  The creationist claim that HBBP1 is not junk is proved, by the strongest kind of evidence, and proved more strongly than Tomkins claimed.  The TSS deletion was earlier read here as falsifying the creationist mechanism, regulatory transcription;  it does so only for gamma-globin silencing in HUDEP-2 cells.  Ma, Liu et al. 2021 (ma-2021b, below) found the RNA essential for red-cell development, so the 2026-10-27 review downgraded "falsified" to "one end point unaffected".  None of it bears on whether the sequence was inherited or installed.

The human beta-globin pseudogene is non-variable and functional

Tomkins, Jeffrey P.  Answers Research Journal 6:293–301, 2013.

camp: creationist topic: hbbp1 method: measurement direction: complicates verified: full

Key figures: HBBP1 about 2,350 bases, 97.8% identical to chimpanzee, 97.8% to gorilla, 92.2% to orangutan, 92.1% to macaque;  16 exon variant transcripts, 42 intron variant transcripts, expression in 251 cell types.

The best creationist paper in this catalogue:  right about what it set out to prove, silent about what matters.  On function it is vindicated by Moleirinho, Huang and Ma 2021b.  On mechanism it is mostly vindicated by Ma 2021b, which found the HBBP1 RNA essential for human red-cell development;  Huang's TSS deletion left gamma-globin silencing unaffected, and the breadth claim (251 cell types) does not fit Ma's bone-marrow-only expression.  On ancestry it does not argue at all, offering one sentence of "code re-use" with no prediction and no number.  Its own data contain a perfectly nested primate series that it does not comment on.  The transcript counts are Tomkins's reading of browser tracks with no independent published source and are not quoted in the article.

Genome-wide analysis of pseudogenes reveals HBBP1's human-specific essentiality in erythropoiesis and implication in β-thalassemia

Ma, Yanni, Siqi Liu, Jie Gao et al., Yong E. Zhang & Jia Yu.  Developmental Cell 56(4):478–493.e11, 2021.  https://doi.org/10.1016/j.devcel.2020.12.019  Library key ma-2021b.  Erratum:  Dev Cell 60(12):1807, 2025, doi:10.1016/j.devcel.2025.05.021.

camp: mainstream topic: hbbp1 method: measurement direction: complicates verified: full

Key figures: p. 481 — HBBP1 RNA expressed "exclusively in bone marrow", above 98.9% of protein-coding genes on day 8 of erythroid differentiation;  whole-locus or 260-bp TSS deletion in hESCs cuts BFU-E colonies from about 58 to about 2;  shRNA knockdown to 20-30% impairs maturation in cord-blood and adult bone-marrow progenitors and in a xenograft.  p. 482 — HUDEP-2 splice-site mutants that remove mature RNA lower globin and hemoglobin;  exogenous RNA partly rescues knockout cells.  p. 486 — HBBP1 "not or poorly expressed" in chimpanzee blood (RNA-seq plus qRT-PCR on five more chimpanzees);  authors infer human-specific co-option.

The paper that turns the HBBP1 transcript question.  The library held its abstract only, recorded as walled;  the full 48-page publisher PDF came from the Wayback snapshot 20221116234121 of the cell.com PDF on 2026-10-27 (cell.com and ScienceDirect return Cloudflare 403).  Pages anchored on the printed footers, PDF p5 = 481 and PDF p10 = 486.  Direction is complicates:  it supports the creationist function and RNA-mechanism claim for HBBP1, contradicts Moleirinho's "marginally or not transcribed", sits in tension with Huang's TSS result on a different end point, and reads the human-only function as co-option, which leaves the shared lesions (Chang 1984) as they were.  The 2025 erratum replaced a duplicated RT-PCR gel in Fig. 5A;  the authors say it does not affect the conclusions.

The egg-yolk gene

Loss of egg yolk genes in mammals and the origin of lactation and placentation

Brawand, David, Walter Wahli & Henrik Kaessmann.  PLoS Biology 6(3):e63, 2008.  https://doi.org/10.1371/journal.pbio.0060063

camp: mainstream topic: vitellogenin method: measurement direction: undercuts verified: full

Key figures: VIT remnants in regions syntenic with the chicken genes;  VIT1 alignment scores significantly above the genomic background by a Mann-Whitney U test;  shared indels in VIT1 exon 3 between human and dog, and two more shared with armadillo.

The source of the whole vitellogenin argument, and it does not rest on one 150-base fragment.  Its load-bearing evidence is the shared indels, which Tomkins never mentions.  The authors print their own cautions, which the creationist side should be quoting instead:  the VIT2 and VIT3 matches "may be too short to provide statistical significance or partially spurious", and VIT genes were independently inactivated more than once in mammals.  The supplementary alignment carrying the indels was not inspected here.  CC BY.

Challenging the BioLogos claim that a vitellogenin pseudogene exists in the human genome

Tomkins, Jeffrey P.  Answers Research Journal 8:403–411, 2015.

camp: creationist topic: vitellogenin method: measurement direction: undercuts verified: full

Key figures: the alleged remnant a single 150-base sequence at hg19 chr1:79,254,632–79,254,781, 62% identical to chicken over that stretch and 39% over 36,450 bases;  one shared flanking gene in ~360,000 bases;  a Genscan-predicted six-exon "GAM" gene spanning the fragment.

Attacks a leg the argument does not stand on.  Tomkins never mentions indels, dog, armadillo, or the Mann-Whitney test, so a reader of this paper alone would not know the shared-indel argument exists.  Both of his own pillars are weak:  Genscan predicts protein-coding genes and is here used as evidence for a long non-coding RNA, and the claim that GAM genes inhibit translation of target genes cites, in his own reference list, United States Patent 7,250,496.  That is the single most quotable defect in the creationist genetics literature reviewed for this article.  His synteny result, one conserved flanking gene, is the nearest neighbour and is a point for Brawand.

Tomkins on the human vitellogenin pseudogene: who needs signal when you have noise?

"RM".  EvoGrad, 30 July 2019.  https://evograd.wordpress.com/2019/07/30/tomkins-on-the-human-vitellogenin-pseudogene-who-needs-signal-when-you-have-noise/

camp: mainstream topic: vitellogenin method: commentary direction: undercuts verified: full

Key claims: the mammalian VTG remnant is about five sequences totalling ~3,100 bp, not 150 bp;  ten genes upstream and ten downstream shared between human and chicken apart from an IFI44/IFI44L insertion;  the Bentwich "GAM" sequences appear all over the genome;  the Maunakea brain data Tomkins cites is a single read at a locus whose own paper required at least five.

The fullest published rebuttal of Tomkins on vitellogenin, and its author is pseudonymous, so it is cited as a secondary source and labelled as one.  Its claims are checkable at the UCSC browser and were not independently re-checked here.  License unknown, excerpt only.

Both patterns in one table

Human specific loss of olfactory receptor genes

Gilad, Yoav, Orna Man, Svante Pääbo & Doron Lancet.  PNAS 100(6):3324–3327, 2003.  https://doi.org/10.1073/pnas.0535697100

camp: mainstream topic: olfactory method: measurement direction: complicates verified: full

Key figures: 50 OR loci sequenced in five primates with the position and nature of every disruption;  human OR disruption rate 3.3 times the phylogenetic mean;  54% of sampled loci pseudogenised in human against 32% chimpanzee, 28% gorilla, 32% orangutan, 36% rhesus.

The best single test in this catalogue, because one lab in one experiment produced both patterns.  A tally made for this article from Table 1:  of 26 loci disrupted in two or more species, 11 carry the identical lesion at the identical nucleotide and 15 carry different lesions at different positions;  human against chimpanzee, 15 of 18 identical;  and OR2Q1p carries three separate deletions at 505, 507 and 518 in three species.  Nobody on either side of the creation debate appears to have used this table.  The tally was read off the printed table and has not been re-derived from GenBank accessions AY192803–AY192982.

The creationist model, at its fullest

The natural history of retroviruses: exogenization vs. endogenization

Liu, Yingguang & Charles Soper.  Answers Research Journal 2:97–106, 2009.  https://answersingenesis.org/genetics/the-natural-history-of-retroviruses/

camp: creationist topic: erv-count method: hypothesis direction: complicates verified: full

Key claims: ERVs were created as part of the host genome and exogenous retroviruses derive from them;  orthologous ERVs "were created to occupy similar genomic loci in separate species by a single designer";  directed integration may partly explain positional conservation;  syncytin-1 comparative genomics "especially challenging to creationists".

The creationist case at its best, and its best is qualitative.  Liu marshals the function evidence well, the exogenization hypothesis is not absurd and is entertained in the mainstream literature, and he names his own hardest problem instead of hiding it, which deserves credit and quotation.  What it does not contain is a number.  "Created to occupy similar genomic loci" restates the observation and makes no prediction about position:  no expected count, no account of which loci should match, no explanation of the 361 that do not, and no reason the matches should nest by node.  The paper does carry a "Predictions and Expectations" section (more ERV-host functions, DNA-repair roles, restorable degenerate functions, gene transfer between cells), and its "Beneficial functions of ERVs" section gives env three host jobs (fusion, maternal tolerance, receptor interference).  The 2026-10-27 review cites both:  the first to correct the article's "no prediction that could fail", the second to correct its claim that a design account leaves env with no job.  His "directed integration" appeal jumps from yeast Ty targeting, which is tight, to primate ERVs, where the measured targeting is 1.3- to 1.7-fold.

Identification and analysis of unitary pseudogenes

Zhang, Zhengdong D., Adam Frankish, Toby Hunt, Jennifer Harrow & Mark Gerstein.  Genome Biology 11:R26, 2010.  https://doi.org/10.1186/gb-2010-11-3-r26

camp: mainstream topic: background method: measurement direction: complicates verified: abstract

Key figures: 76 unitary pseudogenes disabled "at a fairly uniform rate throughout primate evolution";  11 polymorphic in living humans;  two are pseudogenes in non-human primates and appear resurrected in the human lineage.

Considered and used only as background.  Its honest caution — pseudogenisation is a live, ongoing, sometimes reversible process rather than a fossil record laid down once — is folded into the article's treatment of independent loss rather than cited separately.  CC BY.

The 'VIGE-first hypothesis'—how easy it is to swap cause and effect

Terborg, Peer (the printed byline;  library key borger-2013).  Journal of Creation 27(3):105–112, 2013.  https://creation.com/images/pdfs/tj/j27_3/j27_3_105-112.pdf

camp: creationist topic: syncytin-assembly method: argument direction: complicates verified: full

Key figures: p. 108 — "at least six independent gene-capture events by mammalian genomes";  the assembly of the primate syncytin-1 locus needs "a great number of highly unlikely genetic events of which the selective value of individual arrangements remains highly doubtful".

The fullest creationist objection to the syncytin story, and an argument about mechanism rather than position.  Obtained 2026-09-06 with plain curl from creation.com;  no wall.  It is cited in the article for three quotations, the third of which cuts against its own author's case:  Terborg reports that the syncytins "map to different genomic locations", which is exactly the non-syntenic signature the article uses to show what independent capture looks like.  The improbability argument itself is not answered here, because it does not bear on position;  it belongs in an article on the origin of new genes.

The chimpanzee genome and the problem of biological similarity

Wood, Todd Charles.  Occasional Papers of the Baraminology Study Group 7:1–18, 2006.  Library key wood-2006.

camp: creationist topic: background method: review direction: undercuts verified: full

Key figures: p. 9 — "it is difficult to explain why pseudogenes with the exact same substitutions or deletions would be shared between species that did not share a common ancestor."

A young-earth creationist biologist's review of the chimpanzee genome, written for creationists.  Added 2026-10-27 to keep this article consistent with Shared Genetic Errors, which already cites the same sentence.  Page anchored on the printed folios (PDF p10 = 8, p11 = 9, p12 = 10).  Cited in the article's "Where this article's own side is weakest" as a creationist's own statement of the problem.

Integrated regulation of class II human endogenous retroviruses in a breast cancer cell line

Liu, Yingguang & Tam D. Nguyen.  Proceedings of the International Conference on Creationism 8, article 16, pp. 191–199, 2018.  https://doi.org/10.15385/jpicc.2018.8.1.19

camp: creationist topic: erv-position-model method: measurement direction: complicates verified: full

Key figures: p. 198 — "Structural and positional homology of HERVs between humans and animals are more likely due to functional necessity rather than common descent."

Obtained 2026-09-06 through the Wayback Machine after digitalcommons.cedarville.edu went dead;  the DOI now resolves to cedarville.tind.io.  This was the most likely place a quantitative creationist account of orthologous ERV positions would have appeared, nine years after Liu 2009.  It contains none:  no expected count, no probability, no predicted fraction of shared loci.  The paper is a laboratory study of HERV regulation by methylation, progesterone and OCT4, and its one sentence on shared position is a restatement.  Cited in the article for exactly that, because the absence is the finding.

Inactivation dates of the human and guinea pig vitamin C genes

Lachapelle, Marc Y. & Guy Drouin.  Genetica 139(2):199–207, 2011.  https://doi.org/10.1007/s10709-010-9537-x

camp: mainstream topic: gulo method: measurement direction: undercuts verified: abstract

Key figures: inactivation dates of 61 MYA (primate) and 14 MYA (guinea pig);  the transposable-element analysis "did not reveal conclusive evidence that illegitimate recombination between repeats has contributed to the loss of exons".

Paywalled at Springer;  the abstract was read as bytes from the Europe PMC REST API on 2026-09-06.  This is the paper Tomkins 2014 draws on to argue that the GULO exon losses "occurred independently in each taxon".  It reports a single primate inactivation event and a null result on the recombination mechanism, which is the opposite reading.  Cited in the article for that reversal only.  No breakpoint coordinate or transposable-element identity from this paper is printed anywhere, because the body has not been read.

Isolation and nucleotide sequence analysis of the beta-type globin pseudogene from human, gorilla and chimpanzee

Chang, Louise Y. E. & Jerry L. Slightom.  Journal of Molecular Biology 180(4):767–784, 1984.  https://pubmed.ncbi.nlm.nih.gov/6098690/

camp: mainstream topic: hbbp1 method: measurement direction: undercuts verified: abstract

Key figures: p. 767 — the three hominoid pseudogenes differ by "only about one-third of that predicted for DNA sites evolving at the neutral rate";  and they "each share the same substitutions in the initiator codon (ATG----GTA), a substitution in codon 15 which generates a termination signal TGG----TGA, nucleotide deletion in codon 20 and the resulting frame shift".

The origin of the conservation figure the whole HBBP1 argument descends from.  Abstract read as bytes from Europe PMC on 2026-09-06, cross-checked against PubMed;  the body is still closed at Elsevier.  Tomkins quotes the first sentence accurately and at the right page.  He does not quote the second, which is three identical disabling lesions at three identical codons in three species and is the strongest single piece of shared-lesion evidence in this article.

Long noncoding RNA HBBP1 enhances γ-globin expression through the ETS transcription factor ELK1

Ma, Shu-Ping et al.  Biochemical and Biophysical Research Communications 552:157–163, 2021.  https://doi.org/10.1016/j.bbrc.2021.03.051

camp: mainstream topic: hbbp1 method: experiment direction: complicates verified: abstract

Key figures: overexpressing HBBP1 raises γ-globin in HUDEP-2 cells;  ELK1 binds the HBBP1 promoter;  and "Surprisingly, ELK1 also directly bound to and activated the γ-globin proximal promoter."

Closed at Elsevier;  abstract read as bytes from Europe PMC on 2026-09-06.  The article cites it as the live objection to treating Huang 2017 as final, and then states why it does not overturn it:  this is gain-of-function in a cell line, not a test at the endogenous locus, and the authors themselves report that ELK1 acts on γ-globin directly, which breaks the causal chain the result would need.  A stronger transcript-function claim exists in Ma et al. 2021, Developmental Cell 56(4):478–493, which was not obtained and should be chased before this section is revised.

Loss of egg yolk genes in mammals — supplementary Figures S1 and S2

Brawand, David, Walter Wahli & Henrik Kaessmann.  PLoS Biology 6(3):e63, 2008, supplementary material.  https://doi.org/10.1371/journal.pbio.0060063

camp: mainstream topic: vitellogenin method: measurement direction: undercuts verified: full

Key figures: Figure S2, the VIT1 exon 3 alignment — 152 columns, three gap columns at 14, 51 and 71;  human and dog share all three, armadillo shares two.

Obtained 2026-09-06 from https://www.ebi.ac.uk/europepmc/webservices/rest/PMC2267819/supplementaryFiles;  the PLoS .sNNN endpoints 404.  Figure S1 is a schematic with no sequence;  Figure S2 holds the alignment, as live text rather than an image.  This closes the one gap the article's vitellogenin section had:  it can now print the shared-indel count instead of asserting that shared indels exist.  CC BY.

Considered and not used

• Zhang 2010 and Drouin 2011 are listed above with their reasons: both are cited in the article only by reference to Shared Genetic Errors, where their material already appears.
• Belshaw 2007 is cited for one sentence in the objections and nothing more, because the ERV-deletion argument belongs to Shared Genetic Errors.
• Go, Yasuhiro & Yoshihito Niimura, Molecular Biology and Evolution 25(9):1897–1907, 2008, on human versus chimpanzee OR repertoires. Closed at Oxford Academic; bin/cdp-fetch.py failed. Not load-bearing, since Gilad 2003 was obtained in full.

Could not be obtained

Four of the five entries that stood here on 2026-09-06 were resolved that day and have moved into the body of this file.  What is left is the body text of two papers whose abstracts were obtained, and one paper never needed.

• Lachapelle & Drouin 2011, Genetica 139(2):199–207 — body only. The abstract was read from Europe PMC and is enough to show that Tomkins reverses the paper's conclusion. The body is paywalled at Springer; the wall is payment, not bot-blocking, so bin/cdp-fetch.py renders the landing page and stops there. ResearchGate's publication endpoint is DataDome-blocked and the uOttawa repository (ruor.uottawa.ca) was returning 503, which is the best untried route. An institutional login or a copy from Guy Drouin would open it. Nothing in the article depends on the body, and no breakpoint coordinate or transposable-element identity from it is printed.
• Ma et al. 2021, BBRC 552:157–163 — body only. Note the corrected volume and pages; this file previously recorded them as 565:8–14, which was wrong. The abstract names the confound in the paper's own causal chain, which is what the article uses. The knockdown data have not been seen. A stronger claim in Ma et al. 2021, Developmental Cell 56(4):478–493, was identified and not obtained; cell.com 403s every route tried.
• Go, Yasuhiro & Yoshihito Niimura, Molecular Biology and Evolution 25(9):1897–1907, 2008, on human versus chimpanzee OR repertoires. Closed at Oxford Academic; bin/cdp-fetch.py failed. Not load-bearing, since Gilad 2003 was obtained in full and re-tallied for this article.