T_F family statistics

Divergence

• There are enough variants to account for an average age of 94 Kyr, but with a high variance.
• In coding sequences, the density of private substitutions is 0.002/bp.
• This drops to 0.0014 in the region 6110..6700.
• In 3' UTR the density is 0.0025
• Among the 14 full length coding sequences the likelihood of a coincidental shared substitution is .002 x .002 x 5000 x 14 x 13 = 3.6. For a coincidantal confirming substitution it's 3.6 x .002 x .002 x 5000 = 0.07. Hence a grouping enforced by 1 position is not significant; 2 positions gives 93% confidence and 3 or more gives very high confidence.
• If a 3rd sequence agrees, the liklihood of the double coincidence is 3.6 x .002 = 0.007. So even a single shared variant by 3 or more sequences indicates common ancestry with high confidence.
• Among 29 coding sequences in range 6110..6700, expectation of one coincidental shared variant is .0014 x .0014 x 590 x 29 x 28 = 0.9. Likelihood of corroboration by a second position is 0.9 x .0014 x .0014 x 590 = 0.001. In this case 2 positions gives a high confidence.
• The 3' UTR falls in the same range.
• There are 3 instances of only 2 sequences sharing a solitary informative position; these are likely to be coincidences. All other shared variants probably indicate common ancestry. To the extent that these positions are homoplasic, they indicate recombination or gene conversion.

Phylogenetic subdivisions

Three subdivisions of the TF family are distinguishable: indicated as TF1, TF2 and the L1spa group. Six differences between The differences distinguishing TF1 and TF2 are concentrated between positions 6845..6955. This almost certainly reflects a small patch of gene converted sequence. 5/6 of the differences are reversions to an ancestral state found higher up on the tree, consistent with this being a gene conversion from some older sequence. The tree shown is obtained by forcing these 6 variants to covert simultaneously and fixing the direction by making the ancestral state the one found in spretus. Sequences were placed relative to their content of this converted region, with disagreeing positions left to be explained as coincidences or recombinants.

• L1spa group: The grouping of L1spa and AF081109 is supported by a 100 bootstrap. There is also support for the affinity of these two sequences in the promoter array. AF081111 fits this group in the right part of the sequence, but fits with TF1 in the left part. AF081111 is therefore considered to be a recombinant and L1spa can be considered as a representative of a lineage that is developing separately from the others. Dad will also require a conversion or recombination.

• TF1 group: The TF1 group splits further to a subgroup AF081104,AF081105, and subgroup AF081114,AF081108,AF081110. The latter acquired the insertion at 7268 in parallel with the L1spa group. AF081110 has 4501 T, apparently acquired by conversion from the other subgroup, and this pattern is corroborated by the left half of AF081111. The distribution of 6226G also requires a conversion. The strong affinity between AF081104 and AF081105 collapses further to the left in the sequence with AF081105 becoming a very close match to the L1spa group in the promoter array. However, the remainder of the TF1 group has weak affinity within the promoter array, allowing identification of a full length consensus representing this point in time. The best full length representative is AF081104.

• TFnode2: The remainder of the sequences fall in an undifferentiated burst indicated as TFnode2. TFnode2 is best represented by L1Orl. L1Md4 likely acquired 6226 C from TF1 by conversion.

Conversion patterns

• From outside TF

  AF081108 is gene converted to something more A2-like in a patch from 4486..5304 covering 11 informative positions. There is a potential patch of gene conversion from an A2-like sequence in AF081107 covering 3 informative positions between 3779 and 3796. There are 2 potential 2 dinucleotide events. and 7 potential single nucleotide matches some of which might represent gene conversion from an A2-like sequence.

  • There are 65/2970 bp differences between A2 and TF in the coding region after the recombination point (3733). = .022/bp
  • The expected number of singlenucleotide identities between A2 and a TF long element is .022 * .002 * 14 * 2970 = 1.8
  • There are 23 total agreements with A2 in this region, including the 11 bp obvious conversion. There are an average of 10 differences/TF sequence, so the expectation of another A2 match next to the last is: 2970 * .022 * .002 * 1/5 * 1.8 = .04, so every time a 2 consecutive match comes up, it has a 95% confidence of being a conversion, as long as there are not hypervariables involved.
  • The number of single nucleotide matches to A2 exceeds the expectation although rejection of the statement that they are all coincidences carries only about 93% confidence.
  • The double and triple nucleotide tracts involve very closely spaced positions. The tract size would usually have to be less than 50 bp to avoid more multiposition transfers of more widely scattered bases.
  • Conversion by A2 accounts for no more than 5% of the single nucleotide variation in the TF family. The amount coming from other sources is not yet measured.
  • The number of conversion events is distributed unevenly, with as many as 4 affecting one sequence, while half of the full length sequences show no sign of conversion.

• Within TF

  Only about a half dozen single nucleotide homoplasies are minimally required to satisfy the tree. This would set a relatively low limit on within family single nucleotide conversion. The two putative intra family recombinants, AF081111 and AF081105 can not be distinguished from multinucleotide conversion affecting the ends, since the TSD regions were not reported.

  If tracts were similarly less than 50 bp between TF elements, then less than 10% of conversions would be detected (because no differences would be transferred). Yet the total amount of conversion detected within TF and between A2 and TF is about the same. The number of copies of the two families is not radically different. Therefore, conversion within TF is apparently more efficient, either in terms of more events or longer tracts.

Recombinants

Nonsynonymouse:synonymous tabulation

The lineage before the burst

Promoter arrays

Sequence Sources

• AF081104 to AF081114
• L1Spa
• DAD,DAE,DAA,DAB,DAN,DAM,DAO,DAP,DAQ DAS, DAR are from DMZLW40
• SAF from SPLW39
• PA, PD from PANLW61
• Musmht
• L1Orl
• AC4405lr
• D2,3
• AC3949LR
• Musrsl1r
• AC5403lr
• L1Md4

  Last major revision 5/10/99; Last update 2/16/00