REV pol (protease and reverse transcriptase) forward

Sequence (5′ to 3′)
CAAATAATAGATTTTCTAGTAGATACGGGA
Reverse complement (5′ to 3′)
TCCCGTATCTACTAGAAAATCTATTATTTG
Direction
forward
Target
REV pol
Set
PCR REV pol (protease and reverse transcriptase)
Pairs with
REV pol (protease and reverse transcriptase) reverse
Length
30 nt
GC content
30%
Melting temperature estimate (Tm)
54.9 °C
Annealing temperature
Depends on the polymerase: NEB Tm Calculator
Used in

The melting temperature (Tm) is an estimate for comparing primers with each other, computed from the sequence with the SantaLucia (1998) unified nearest-neighbour model (SantaLucia J Jr (1998), Proc Natl Acad Sci USA 95:1460-1465) at 50 mM monovalent cation and 0.5 µM primer. These are standard reporting conditions, not those of any one reaction buffer. A sequence with an ambiguity code has no single Tm and shows none. The Tm estimate is not an annealing temperature. For the annealing temperature of a particular polymerase, use the NEB Tm Calculator.

In the literature

Published in
Not found: Stewart et al. 2019 does not print this primer. It cites Barbosa et al. 2007 (Virus Res 124:68-77, doi 10.1016/j.virusres.2006.10.002) for the REV pol segments; that paper and Stewart et al.'s supplement could not be read (publisher access blocked).
Name in the paper
Not found: The paper that prints it could not be read (see published_in).
Sequence in the paper
Not found: The paper that prints it could not be read (see published_in).
Product in the paper
Not found: The paper that prints it could not be read (see published_in).

On the reference

Reference
DQ387450.1: Reticuloendotheliosis virus strain APC-566, complete genome
Position
  • 2492 to 2521, + strand, no mismatches
Product with REV pol (protease and reverse transcriptase) reverse
  • 574 bp, 2492 to 3065

Positions and product sizes are computed from the sequence and the reference by the site’s own tested primer code, in the accession’s own coordinates, and are not copied from a paper. A primer is looked for exactly first, then with more mismatches up to 4, and any mismatch is shown.