An engineering paradigm was used to predict the logical sequence of events that leading to the origin of the current genome where informational and catalytic properties are separated in two entities. This hypothesis can be tested by investigating whether the proposed functional steps can be translated into a mechanistic sequence of events. In this section, a gradual scenario for a double-stranded RNA origin of life is proposed that reflects the proposed functional scenario that was based on design-by-contract.
From RNA oligomers to dsRNA
The 'dsRNA first' hypothesis about the origin of life states that the first functional step was the appearance of dsRNA before the development of catalytic RNA and thus implies that dsRNA appeared abiotically. Assuming that short strands of RNA formed by template-directed abiotic ligation were available in a prebiotic world (for discussion see [16,17], dsRNA can be formed by the hybridization and ligation of complementary sequences of oligonucleotides (Fig. 2). In this process, oligonucleotides could initially built up an interrupted double-stranded chain of RNA, which was followed by a non-enzymatic ligation reaction to form an uninterrupted double-stranded piece of RNA. This mechanism is similar to the template-directed, non-enzymatic ligation and amplification of oligonucleotides [3,18-23] proposed to replicate an existing catalytic RNA molecule. Thus, the template-directed hybridization and extension of short oligonucleotides is a feasible approach for abiotic formation of longer strands of dsRNA.
Replication by an abiotic chain reaction
The crucial step in the origin of life is the formation of a replicating system that will allow genetic information to be transduced and amplified. The replication of dsRNA can be based on the intrinsic properties of complementary strands of ribonucleotides to form a stable double-stranded helix below the melting temperature of dsRNA and to separate into individual strands at temperatures above its melting temperature [24,25]. After temperature-induced strand separation, oligonucleotides can rehybridize to both individual strands upon lowering of the temperature to form new chains of interrupted dsRNA (Fig. 3). After non-enzymatic ligation of the interrupted strands (cf. Fig. 2), two new strands of dsRNA are formed. These newly formed dsRNA strands can then re-enter melting and rehybridization cycles, thereby amplifying the original strands of dsRNA. This process is similar to the polymerase chain reaction (PCR) used to amplify dsDNA, albeit with much slower polymerization time and could be driven by the diurnal cycle (see Discussion). The hybridization and extension of primer oligonucleotides to each other is frequently seen in PCR (e.g. [26-29] showing that thermal cycling in combination with ligation reaction can elongate as well as replicate oligonucleotides in vitro. Thus, using only non-catalytic short strands of ssRNA, dsRNA as an informational molecule can be formed and replicated abiotically.
An early transcription mechanism
The next step after the establishment of a pool of replicating dsRNA as a genetic information carrier is the subsequent generation of ssRNA that could function as a ribozyme (catalytic RNA; cf. Fig. 1). This generation of ssRNA from dsRNA can be accomplished by the partial melting of the double-stranded helix, followed by the hybridization of RNA oligonucleotides to the resulting partially separated strand (Fig. 4). The hybridized oligonucleotides can then be connected by a slow abiotic ligation process, basically similar to the one that is proposed in figures 2 and 3 in the generation and replication of dsRNA. This ssRNA can also be elongated in subsequent cycles since the partial sequence will selectively rehybridize with its anti-sense RNA, leading to a full size' transcript' in multiple ligation cycles. A subsequent release and folding of the hybridized RNA sequence from the template strand at higher temperatures would produce the first catalytic RNA (or later mRNA), a process that uses the same interface and is conceptually similar to enzyme-assisted transcription.
A change in temperature and/or salinity may cause a partial melting in regions of the dsRNA where the local melting temperature is lower due to the base composition. This partial melting or 'breathing' is a well-described phenomenon in dsDNA that occurs when thermal fluctuation opens up part of the dsDNA (passive opening) and allows components of the transcription machinery to bind the exposed single-stranded DNA [30,31]. Regions of RNA with specific nucleotide sequences (e.g. AU-rich) that have low local melting temperatures are more subject to partial melting and these specific regions could have represented early genes. Partial transcripts that rehybridize with partially melted RNA may also prevent the helix to close and thereby enhance their own transcription. Thus, the formation of an abiotic transcription bubble and subsequent abiotic transcription can be based on physico-chemical characteristics of dsRNA.
From a dsRNA to a dsDNA world
The advent of protein generation (translation) can speed up evolution by replacing existing abiotic replication and transcription processes by more efficient protein enzymes, for instance RNA polymerases. Based on the presence of existing dsRNA, in only a few gradual changes in an existing RdRp, the transition from a dsRNA to a dsDNA world can be made . This transition is conceptually simple because it can be made by substituting the ribonucleotides building blocks of RNA with the deoxynucleotides of DNA (cf. Fig. 1). The insertion of deoxynucleotides by turning an existing RNA-dependent RNA polymerase (RdRp) into an RNA-dependent DNA polymerase RdDp) would effectively create a DNA-RNA hybrid, while the concomitant evolution to a DNA-dependent DNA polymerase (DdDp) would create dsDNA (Fig. 5A). The existing transcription process would only need a change in the template recognition site of an RdRp to a DdRp polymerase (Fig. 5B). Substrate specificity can be influenced by subtle modifications to a generic polymerase module  and most DNA polymerases are able to incorporate rNTPs as a substrate instead of dNTPs [33-36]. Also, template recognition of polymerases is not very specific and can be changed by minor mutations [37-40]. This would not have affected existing replication mechanism or involved the de novo development of a transcription machinery.