Why not just use GenBank?
Quick quality checks of new prion sequences
Putting the database to use: examples
High priority sequencing needs
Common mistakes in prion genomics
Format and structure of the sequence data
Getting exactly the data you need
Database update log
25 Jan 00 webmasterThis page is gateway to a complete curated database of all available prion and prion-like sequences. While encompassing all 497 prion entries at GenBank, the database additionally contains unpublished and never submitted sequences (based on review of full text and secondary citation). It also includes sequences reliably inferrable from blast searches of finished and unfinished genomic, protein and EST databases. Many researchers have cooperated via private correspondence to clarify sequence data, verify or withdraw anomalies and release 'odds and ends' for which there was no space in journal articles.
Just as importantly, entries are curated for probable sequencing error; secondary structure in the hairpin C stem-loop upstream of the first repeat has led to numerous unacknowledged sequence errors. If a given sequence at GenBank seems to be missing, it is because something is very wrong with it. Curatation uses the database itself, secondary and tertiary structural information, reported experimental methodology, and accrued experience with many unrelated genes and proteins to flag and downweight problematic entries (some still informative regionally). Errors at GenBank are forever -- by policy they cannot be corrected except by the submitting laboratory.
Curation is extremely important in a protein with many conserved residues because bad data can dominate apparent rates of evolutionary change, suggest bogus tolerated substitutions, mislead as to normal function of residues, and distort inference of species tree topologies.
The database also contains all known alleles, mutations, and polymorphisms in all species; few of these are to be found at GenBank, Medline, OMIM, Cardiff, or SwissProt. Generally, all point mutations in a given species are 'bundled' here into a single database entry to facilitate site comparison. Point mutations, octapeptide deletions and insertions, and upstream and downstream non-coding regions are dealt with at greater length elsewhere, as are major topics such as the significance of sheep and cow alleles in disease.
The prion gene superfamily on 25 Jan 00 consists of prion gene orthologues (in mammals and birds) plus doppel paralogues (tandem duplicatation) known in mouse, rat, and human. No pseudogenes nor syntenic duplications can be detected in the 48.7% of the human genome sequence released to date. The superfamily is orphaned -- no connection can been made to any zebrafish, drosphila, nematode, yeast, or bacterial gene despite a rather slow rate of evolutionary change over 310 million years. Unrelated genes in yeast are sometimes called prions; these have no homology whatsoever to the vertebrate prion gene family and are not included in the database here. Dozens of further unrelated human gene products also form congophilic cross-beta important in a variety of diseases.
A lab with data not in this database is encouraged to submit it; inclusion does not jeopardize subsequent print publication in any way. Be sure to accurately describe the specie and subspecies, its provenance, the number of individual animal lineages sequenced, and the sequencing method.
One important use of this database is to allow researchers to run real-time quality checks on new sequences. The idea is to give special experimental attention to validating novel or unexpected variation while raw sequenator output and primer templates are still at hand. Unexpected substitutions can and dooccur -- but its submission to public databases is so much more valuable when the researcher is absolutely sure of them. (There is a danger as well -- that someone will interpret questionable reads as the conventional expected value and thus miss changes. It is also possible to use this database or GenBank to construct completely fraudulent, but highly plausible, sequences; one lab apparently fabricated two GenBank entries using published sequences as a point of departure.)
Redundant protein and DNA sequences are retained in the database if they pertain to different species. If nothing has changed in 5 million years since speciation, this is important to know. However, a database field flags redundant sequences, allowing a non-redundant set to be extracted. This reduce distortions from over-sampling of certain sub-trees and reduces visual clutter in comparisons.
Some groups, such as rodents or old world primates, can be economically replaced by a consensus (or ancestral) sequence and a single alternate sequence containing all signficant variation. Similarly, all alleles within a single species can often be consolidated into a single super-allele. This artifice is useful in quickly evaluating new sequences for interesting features and anomalies. As the number of sequence has grown to over a hundred, it becomes necessary to find such compromises to allow a full set to be displayed on a single 19" screen.
Many researchers do not realize just how interesting their data is to evolution and genomics . It is not possible to anticipate future issues in prion research either. For these reasons, it is important to nail down every practible detail of a sequence and make it available in print, GenBank, or lab web site. Example: a 4-repeat allele was noticed in gorilla. This did not seem important at the time. Today we know that these examples are important in identifying the mechanism of repeat region mutation. The gels and sequenator output could not be located years later and important data on the particular repeat deleted was lost.
Optimally the sequencing run should include the intron-exon junction 10 bp upstream along with the complete ORF and some conserve 3' UTR. This does not involve additional primers, time, or expense. Some labs are indeed sequencing short stretches upstream and downstream of the ORF but mistakenly truncate these in submissions to GenBAnk. The two short upstream exons and promoter region could be sequenced in many cases with little additional effort.
11 Nov 99 webmasterHere are some quick questions to ask of a new prion sequence. Rare events do occur (but only rarely). There is nothing wrong with reporting an unusual change if a commensurate extra measure of experimental support is provided and awareness demonstrated.
-- Is the 'nearest neighbor' phylogenetically plausible? Example: a squirrel is a rodent, not a cetacian. Its sequence should align more closely to mouse or hamster than to cow or dolphin. Check in at GenBank taxonomy, then use Blastp at NCBI, setting the filter option to 'none' and output to 'flat without identity' for the sequence with repeats removed. Are the expectation values highest for Rodentia; do rodent-specific codons have their expected value in squirrel?
-- Silent mutations are more commonly fixed than those affecting the amino acid sequence. If the ratio of synonomous to non-synonomous is 0:3 instead of 3:1, how plausible is this? Do the observed amino acid result mainly from single base changes, are transitions in 2:1 ratio to transversions? Are the amino acid substitutions conservative and frequently observed in other proteins (Blosum/PAM matrices)? Cautionary note: a unique substitution requiring two base changes in an otherwise conserved residue was reported in pigs; the lab confirmed this by examining a dozen further animals and additional breeds.
-- Do observed amino acid changes mainly occur at codon positions in the prion gene that show tolerance for frequent change in other species? If 3 changes are observed at positions with 310 million year time scales and 0 changes are observed at positions with 5 million year time scales, that would be surprising indeed. In comparing two closely related species, is the number of changes compatible with an average rate change of 1 change per full length prion protein per 4 million years?
-- Would amino acids changes, because of size or polarity, disrupt the conserved hydrophic core in the three-dimensional structure or are they mainly exterior? Changes in the conserved 'underpass' region are especially improbable in wildtype of any mammal. The region before the first beta strand has unknown 3D structure but is the best conserved domain in the entire protein being identical in all species of bird as well.
12 Nov 99 webmasterPredicting sequences of unknown species. This can be done fairly well provided that the species in question is not a long branch. Some codon positions are only predictable up to a twofold ambiguity at some codons and generally there are some surprises. (In fact, the main value of sequence prediction is in highlighting subsequent surprises.) The species barrier was once thought to be directly proportional to amino acid differences; this is true as a gross rule of thumb but many exceptions prelcude its use in health policy. Indeed, the species barrier is not even symmetric, ie species A may infect species B much more easily than species B can infect A.
Predicting sequences of ancestral species. Many web tools can reconstruct an ancestral sequence. This can be done at various taxonomic grades, eg, bovid, artiodactyl, eutherian mammal, ancestral vertebrate, prion-doppel progenitor. It is not clear how far back fossil DNA can be sequenced in frozen material. Reconstructed sequences help us understand where and when evolutionary change in the prion gene occurred. They serve as better Blast probes for locating relatives in more distant species. These sequences can be threaded onto known 3D nmr structures to recover ancestral 3D appearance and long distance residue relationships.
Evaluating human mutations. One common issue is whether a given observed change is causative for CJD or neutral. In some cases, little or no family history is available. It cannot beee assumed that common polymorphisms such as V129 are inherently neutral (other than modifying disease course in other backgrounds): allele frequency is also high in Mediterrean fever and sickle cell anemia. Silent mutations, if they occur in a region important to mRNA secondary structure and its regulation, are not necessarily neutral either. In some cases, eg N171S, the database suggests that a mutation in human that is normal in other lineages (here chimpanzee) is probably not pathogenic; history of a codon position supplements 3D structure and extensive data elsewhere in evaluating conservativeness of amino acid substitutions.
Evaluating inbred alleles The issues in evaluating livestock and lab animal alleles are similar to those in humans. However, extreme inbreeding has occurred in species such as sheep, apparently inadvertently fixing some rather odd prion alleles along with desired wool or milk traits. Indeed, scrapie in sheep may have arisen centuries ago due in animal husbandry practises. See discussions of mouse oddities and hazardous haplotypes in sheep.
12 Dec 99 webmaster
"Sir, neurologists choosing species to sequence is like a dog's walking on his hind legs. It is not done well; but you are surprised to find it done at all." Samuel Johnson 1763 [Y2K update]
A great deal of hard work led to the set of prion sequences available today. Many are important to prion disease research issues. However, a great deal of wasted effort occurred, attributable to lack of consultation with molecular evolution specialists. This lead to over-sampling of slowly changing lineages, reporting of partial sequences, under-sampling of upstream control regions, non-observance of the triples principle (long branch avoidance), failure to resolve repeat region origin, and orphan status of vertebrate prions (no homologue in early chordates).
Species barrier: Nothing illustrates the bizarre selection of species sequenced to date more than this fact: of 21 zoo species affected by BSE feed, only 5 were ever sequenced. Numerous felid species were affected but no other carnivores: was this a clue not pursued to the species barrier? (Some carnivores were sequenced, but dog-dingo-wolf over-samples as does mink-ferret.)
A cautionary note: poor records were kept on zoo feed and species exposed, a dog experiment was terminated even as spongiform encephalopathy was found in 19 dogs, horizontal transmission in a kudu herd is now attributed to continuing massive leaks in the feed ban [superceding Kirkwood et al. Vet. Rec. 135: 296 1994.], experimental mink succumbed to BSE feed after only 15 months [J Gen Virol 1994 Sep;75 ( Pt 9):2151-5]. This could imply a carnivore cover-up rather than a species barrier (dogs are a highly charged emotional issue). Below are some suggestions for further sequencing (arranged by purpose):
- Contextual sequences for livestock and lab species: Lab animals and domesticated livestock may have fixed unusual alleles, possibly harmful ones, during the course of inbreeding. The prion gene is some strains and breeds may not represent wildtype and even predispose to TSE or already lack normal function. Long incubation mouse is a prime example of this uncertainty. To determine wildtype in these species, it is necessary to sequence contextual species: the nearest neighbors among wild animals.
sheep and goat mouflon Ovis aries musimon mountain goat Oreamnos americanus wild goat Capra aegagrus cow water buffalo Bubalus arnee African buffalo Syncerus caffer nilgai Boselaphus tragocamelus elk and mule deer water deer Hydropotes inermis moose Alces alces caribou Mazama americana pig wart hog Phacochoerus africanus bearded pig Sus barbatus collared peccary Pecari tajacu rodents steppe mouse Mus spicilegus striped hamster Phodopus sungorus bush rat Rattus fuscipes chicken ceylon junglefowl Gallus lafayettei chukar partridge Alectoris chukar silver pheasant Lophura nycthemera- Additionally affected: These species have been used proposed as a risk for human transmission via brain consumption, are commonly co-pastured with scrapie sheep, or eat CWD carrion in winter.
gray squirrel Sciurus carolinens cougar Felis concolor reindeer Rangifer tarandus wolverine Gulo gulo- Leverage existing data: Given the sequences already determined, which additional species best 'add value'? The ildea is to rectify the original choice of species by judicious augmentation. This could dramatically improve structural and functional inference with a small amount of additional effort. For example, isolated species make for long branch distortions; far better to sequences, say, marsupials in threes with known branch structure reliably dated from the fossil record. Exon 1 and 2 could be surprisingly better interpreted if only two further primates and one further artiodactyl were sequenced.
ten residue repeat origin and mammalian prion ancestor: gray kangaroo Macropus giganteus duckbill platypus Ornithorhynchus anatinus hexapeptide repeat origin and ancestral amniote: alligator Alligator mississippiensis spectacled caiman Caiman crocodilus crocodile Crocodylus niloticus cetacian origins chevrotain Tragulus javanicus hippopotamus Hippopotamus amphibius blue whale Balaenoptera musculus- Evolutionary origin of prion gene: The prion gene is evolving very slowly and (like all proteins) has ancient an origin. Salmon are the earliest vertebrate branch confirmed with antibody; KIAA0168 is adjacent to doppel in fish. Clues to normal prion function and aquisition of new structure and function will emerge from tracing the prion gene back to earlier vertebrates, ie, when did the copper-binding repeat region first emerge, how long has the protein had a GPI-anchor and membrane surface location, etc.? It is far more cost efficient to sequence a dozen earlier vertebrates than to host one infected hamster for a year. The species below are chosen for taxonomic coverage, availbility of DNA, and use in prior sequencing projects:
earlier branching vertebrates: amniotes (sauropsids) tuatara Sphenodon punctatus iguana Brachylophus fasciatus gecko Gekko gecko iguana Brachylophus vitiensis snapping turtle Chelydra serpentina snake-necked turtle Phrynops hilarii earlier branching vertebrates: amphibians, lobe-finned, ray-finned, caritlaginous fish leopard frog Rana pipiens spadefoot toad Scaphiopus holbrooki tiger salamander Ambystoma tigrinum coelacanth Latimeria chalumnae marbled lungfish Protopterus aethiopicus bichir Polypterus bichir zebra fish Danio rerio chinook salmon Oncorhynchus tshawytsha spiny dogfish Squalus acanthias earlier chordates and echinoderms: lancelot Branchiostoma floridae tunicate Ciona intestinalis sea lamphrey Petromyzon marinus Pacific hagfish Eptatretus stouti purple sea urchin Strongylocentrotus purpuratus starfish Asterina pectinifera- Further doppel sequences: Doppel is an easily sequenced tandem paralogue of the prion gene lacking both the repeat region and core invariant region 106-126. With just rat, mouse, and human sequences, it is difficult to determine when the prion gene duplicated, how doppel's conserved domains compare, whether doppel is still active in all lineages, and what its role in prion and other disease might be.
most useful initial doppel sequences: hamster sheep, cow, deer, elk pig, llama, mink gorilla, chimp macaque, squirrel monkey, lemur kangaroo, brush-tailed opossum, bandicoot
9 Jan 00Because so many species have been sequenced (here 89), many genomics researchers include the prion gene in their sample of eucaryotic nuclear genes. The prion gene has evolved slowly and has modest phylogenetic signal pertaining to earlier ordinal mammalian divergences. There are few insertions or deletions and many conserved anchors that validate alignments. The ten most common mistakes in the prion genomics literature are:
...(1) to use a small subset of the available prion sequence data, typically only a third, which can severely impact outgroup long branch effects (ie, using 7 bird species is a vast improvement over inbred chicken alone as an outgroup to mammals). Early sequences were too concentrated within old world primates, poorly scattered among genera, and barely exhibited variation at the protein level.
...(2) to include the tandem repeat region in alignments, despite frequent over-writing via octapeptide deletion and re-insertion and thus lack of homology, further complicated by single insertions and deletions in the oligo-glycine runs; only the first and last repeats are spared replication slippage and have phylogenetic value.
...(3) to fail to consider effects on rates of DNA change due to a conserved stem-loop region, hairpin C, which may constrain even silent mutations, and due to CpG mutational hotspots, which are known in mammals to have 4x elevated rates of mutational change and to be severely depleted in the prion gene and in mammals overall (outside of promoter islands). Of 31 known point mutations in human prion, 9 are CpG. While fixation is different, opportunity is also significant especially in drift.
...(4) to include obviously faulty sequences at GenBank involving cross-contaminated PCR with suspected 1-2% error rates that especially skew analysis in this highly conserved gene: Cat sequence AF003087 is faulty because of closer resemblance to sheep than carnivores, unlike alternative sequences for these species; kudu sequence X74771 has a high concentration of improbably non-synonymies near hairpin C secondary structure yet elsewhere agrees perfectly with 2 other kudu sequences; M13667 is flawed, a 1986 sequence that shows a deletion AGAA-AGA in the central conserved region; a 3' UTR, M31313 or AJ223072, in sheep showed 82 variatons in 3246 bp compared to D38179 and U67922 which agreed completely; GenBank dog sequence AF022714 is complete nonsense (dog sequence AF042843 itself has odd residues post-signal, GGWNTGGG instead of GGGWNTGG, but the authors have validated this].
...(5) to not de-weight more subtle sequencing errors, highly concentrated (and thus identifiable) within implausible singlets(eg, trp to arg at a residue otherwise invariant for 310 my in a single species out of 103 from a lab with no prior sequencing experience).
...(6) to use domestic or lab animal sequences in which alleles or deleterious mutations fixed by intensive artificial inbreeding escape severe counter-selection found in the wild; these are not at all comparable to bona fide accepted point mutation, as can be seen from analysis of chicken, mouse, hamster, goat, dog, and sheep prion sequences.
...(7) to not recognize toggle codons at which, say, serine and asparagine are alternative alleles (not necessarily with 50-50 weighting) in all sub-lineages and also within any species of sufficient population; these polymorphisms persist through speciation events and can unduly weight long-range phylogenetic analyses. In runs of glycine, 3 bp indels toggle with individual glycines at positions hard to determine with no apparent effect. Toggle residues argue for representing amino acids as 1x21 vectors having (mostly zero) weights distributed over all amino acids.
...(8) to not accommodate the number of individual unrelated animals that a database entry represents; some species are represented by a single animal, in other species (mule deer, sheep, humans, elk) thousands of individual animals have been sequenced, providing approximate allele frequencies. The second individual human gene to be sequenced [U29185, Puckett et al.] was later recognized to represent a 4-repeat allele that only occurs at a 2% frequency.
...(9) to not include deletions as potential synapomorphies. In the prion gene, after discarding high frequency deletions generated structurally by tandem repeats, certain unique events are more reliable delineators of phylogenetic relations than point mutational change. Indels can be resolved unequivocally in some instances as deletions or inserts. There is minimal concern that a deletion of several codons (outside of a tandem repeat setting) will revert millions of years later to its previous sequence.
...(10) to not use common sense in preference to software programs whose appearance of objectivity and convenience conceal grossly erroneous assumptions, ie, that the rate of evolution has been constant along the prion gene (or that synonomous codons for a given amino acid are used equally). The rate is easily seen to vary by a factor of 50, sometimes abruptly in adjacent codons, sometimes by domain or 3D position or distant hydrogen bonding partner, presumbably because of selective pressure. While some deep phylogenetic nodes are indeed controversial, marsupials are not cetaceans, a mouse is not an ungulate, a wolf does not belong with primates, and so on. An approximate tree, with the vast majority of node topologies well-established, quickly reveals which codons are useless (too variable or toggle) for long range phylogenetic prediction. In mammals, bursts of SINE and LINE element insertions, which are not revertible or parallel-reproducible changes, has resolved many issues definitively while "maximal likelihood" methods continue to flounder.
23 Jan 00 webmasterThe database consists of prion and doppel protein and DNA data for 89 species and 36 variants. Sequence data resides within a single flat-file database (matrix) of 231 fields (columns) x 479 sequence records (rows) containing 278,000 characters when saved as a text file. It is best viewed on tandem 19" monitors or software with window panes. This size presents problems in display over the internet and also in file transfer because each visitor has access to different database software. Color-coding helps highlight interesting features (synapomorphies, toggles, singlets, doubtful residues, clades and the like) but it is not technically feasible to transfer across platforms.
Therefore no attempt is made to display the data usefully. Instead, links are given to web pages of raw data, one each for protein and DNA with repeat region broken out separately. Birds and doppel sequences, which are fewer, also have separate pages. Fields within a sequence are separated by tabs, sequences are separated by paragraphs (carriage returns). Visitors may go to the data type they want, select all, and paste into their favorite spreadsheet or database. It may work better to replace tabs with commas before attempting data import.
Mammalian prion sequences are already in alignment (but not to marsupial, bird or doppel). Each amino acid and each corresponding nucleotide triplet is given its own field to facilitate sorting (eg, which variations occur at a given position). Fasta format works poorly because so many of the sequences were incompletely determined at the ends and because of repeat length variaton, non-homology, and low complexity. (However, it is easy to reformat the data in conventional fasta format by search and replace.) Ragged ends in DNA (ends of sequencer runs) were discarded though in some circumstances 2 out-of-register bases determines the amino acid. Gaps are designated by '- as are residues not determined at ends.
The first two rows give numbering systems relative to human and sheep (mouse and hamster could be added). The first 9 columns are allocated to fields such as genus and species, common name, accession numbers, comments, and various indexes used for counting, extracting features, and sorting (eg, the sequences can be ordered by phylogenic association or alphabetically by genus-species or numerically by accession number). Thus a camel sequence looks like:
69 camel Camelus dromedarius Y09760 - wt main dna >340_Cam.dro.....fer.cam.cam:: atg gtg aaa ... 69 camel Camelus dromedarius Y09760 - wt main pro >340_Cam.dro.....fer.cam.cam:: M V K ...
Bird and marsupial prion sequences do not gap or alignment unambiguously across the whole sequence with the placental mammals that comprise the vast majority of sequences. They are archived separately as well as in the main database. Doppel sequences are few and are also provided separately in fasta format.
Ancestral and consensus protein sequences have been determined for some blocks of organisms. While these are theoretical, not actual sequences, in many groups there is little doubt about ancestral values. If some residues are uncertain, a lower case letter is used. If a fair number of residues are uncertain, a second sequence is used that embodies all the alternatives, again with upper and lower case used to designate degree of certainty or significance.
Tips for working with the full database: After downloading the text files and successfully uploading in a spreadsheet or flatfile database on your platform, save and duplicate the file. Always work on a duplicate copy -- it is often easier and faster to extract what you want by deleting out rows and columns not needed for the project at hand. This is a large and complex object of which only a portion is visible at any given time.
New sequences are added to the update log on a weekly basis. After sufficient new data accumulates, the primary database is updated and later the downloadable version. However, visitors can update their database directly from the log without delay by using provided links. Some data is better treated by other approaches and fits poorly within the primary database.
Doppel: 3 species -- mouse, human, human allele protein and DNA; rat protein only.
Bird, marsupial prions: 8 species plus a major chicken allele; 1 marsupial
All sequences: 89 species + 36 variants; prion and doppel sequences from all species.
09 Jul 00 webmasterNew incoming data is listed here by reverse chronological data of reception, processing status, source, and species. GenBank accessions sometimes do not have an associated publication; they can also be revised with no explanation of changes made. The database and most trickle-down products are usually updated within a few days of receiving significant new data. Not all data is suitable for database inclusion; mutation data is better treated as stand-alone compilations.
09 Jul 00 included special Tranulis cow, sheep partial doppel sequences 07 Jul 00 awaiting clarification Raymond improved full length deer, elk, antelope prion alleles 01 Jul 00 not included Zhang pigeon, quail prions; taxonomy mix-up unresolved 06 Apr 00 not included Shiue horse short fragment, AF134232, redundant 26 Mar 00 included special Laplanche 5 human doppel alleles 17 Mar 00 included special Simonic turtle genomic 11 Feb 00 included pt db Laplanche 3 new human point mutation 22 Jan 00 included webmaster updated and proofed 11 Jan 00 included pt db Finckh 2 new human point mutations 11 Jan 00 included pt db Windl 9 new human point mutations 18 Nov 99 included Frederikse guinea pig, Cavia porcellus, flawed fragment AF197164 23 Mar 99 included Schlaepfer cow with 7 repeats 10 Sep 99 included Gilch lemur Varecia variegata fragment AF177293 3 repeats 1 Sep 99 included GenBank,JMB doppel 3 species; additional human allele at Sanger Centre 10 Aug 99 included O'Rourke bighorn sheep, complete Ovis canadensis AF166334 10 Feb 99 included Rubenstein new rabbit allele, extra glycines 08 Feb 99 included Schatzl 27 new mammal and 9 new avian sequences 21 Jan 99 not included Hunter 8 goat alleles, new wildtype, not at GenBank 27 Nov 98 not included Doyle greyhound GGG anomaly confirmed 19 Nov 98 not included webmaster flattened bird and marsupial 19 Nov 98 not included webmaster early mammal, early primates 19 Nov 98 not included webmaster early bird ancestral 18 Nov 98 included webmaster 8 bird, 1 marsupial seq added 12 Nov 98 included various added termination codons, ragged dna ends 12 Nov 98 included webmaster added colored human, mouse numbering 11 Nov 98 included webmaster human: all-mutations 'allele' 11 Nov 98 alert webmaster M13667 found defective: GAAA-G 10 Nov 98 processed Piccardo human D202N,Q212P 9 Nov 98 not included webmaster primate ancestrals 8 Nov 98 not included webmaster primate alternates 7 Nov 98 not included webmaster primate alternates 6 Nov 98 processed email human H187R 6 Nov 98 included email Odocoileus termini 7 Nov 98 included webmaster Cercopithicus diana adjustment 3 Nov 98 included Cervenakova fallow deer, Dama dama 2 Nov 98 included O'Rourke pronghorn F090852 1 Nov 98 included O'Rourke white-tailed deer AF091558-60