Isochores and the draft human genome sequence

Early in 2001. two papers reported draft sequences of the human genome. Both of them dealt with, among other subjects, the broad genomic landscape and gene density. The paper by Venter et al. (2001) summarized these features in a Table (reproduced here as Table 3.5) in which the estimates from the human sequence were compared with our estimates (Zoubak et a ., 1996 Bcrnardi, 2000a). Venter et al. (2001) found difiercnces between the amounts of isochores observed by them and those predicted, i.e., based 

The second paper (Lander et al., 2001 also referred to as International Human Genome Sequencing Consortium, 2001) studied the draft genome sequence to see whether strict isochores could be identified and failed to find any. They concluded that their results rule out a strict notion of isochores as compositionally homogeneous and that isochores do not appear to deserve the prefix iso .  

Since the terminology strict isochores was misleadingly used by the authors to denote sequences that cannot be distinguished from random (uncorrelated) sequences (in which every nucleotide is free to change), their failure to identify in the human genome sequences as homogeneous as random sequences (masquerading as strict isochores ) could have been predicted easily on three accounts. 

for over 40 years, since at least the work of Rolfe and Meselson (1959 see also Elton, 1974), random sequences had been known to be much more homogeneous than the least heterogeneous genomic DNAs, namely bacterial DNAs (viral DNAs are not considered here because, if intact, they are perfectly homogeneous see Fig. 1.6). In turn, bacterial DNAs are much less heterogeneous than mammalian DNAs (even if satellite and minor components are neglected). 

strict isochores cannot exist in any natural DNA (i) because coding sequences are made up of codons, in which the compositions of the three positions are correlated with each other (D Onofrio and Bernardi, 1992) (ii) because non-coding sequences are com-positionally correlated with the coding sequences that they embed (Bernardi et al., 1985b Clay et al., 1996 see Chapter 3 below) and (iii) because interspersed repeats are characterized by their own specific sequences. More detailed discussions of this problem were presented by Clay and Bernardi (2001a,b), Clay et al. (2001) and Clay (2001). 

---