Any effective and useful system of transliteration must replace characters of one alphabet with those of another, on a one-for-one basis. The presumed phonological system underlying the written form must come only as a secondary consideration. Technically, then, it does not really matter what letters are substituted, so long as you are consistent in use, and the correspondence is 1:1. In practice, however, it would be misleading and needlessly confusing to assign Roman letters to runic graphemes in a purely random and haphazard fashion.
In the field of runology, there are two dominant and complimentary conventions for the transliteration of inscriptions. In the transcription of Anglo-Saxon and other Old English texts, raw transliterations appear as s p a c e d text, sometimes delimited by single inverted commas (' '). In the transliteration of runic material of a distinctly Norse or Scandinavian character, raw transliterations appear in bold type. Texts which have been normalised, that is to say, rendered into "standard" Old Norse/Old English orthography and edited such that they are suitable for the non-specialist (e.g. word division where not already present, punctuation, and capitalisation), appear in italics. English translations appear in plain type, delimited by single quotes (' ').
In raw transliterations, parentheses ( ( ) ) represent damaged or uncertain runes, an asterisk (*) unreadable but countable runes, an ellipsis (...) unreadable/missing and uncountable runes, a tie (an arc resembling a wide, curved circumflex, not representable under HTML) bind-runes, and square brackets ( [ ] ) runes now absent or unreadable, but known from an earlier drawing or facsimile.
In translation, square brackets ( [ ] ) are used to denote words not present in the inscription but included that the resulting phrase might be more readily parsed.
However, as with any good set of rules, there are exceptions. Particularly with texts from the British Isles, inscriptions may be difficult to date, or rune forms might prove undiagnostic. In such cases, where there is ambiguity as to whether an inscription is or Norse or Anglo-Saxon, the default is to transliterate in bold type.
Neither of these conventions apply phonetic or phonemic notation, for two reasons. Firstly, it makes the transliteration more accessible to the non-specialist, and since runic inscriptions frequently have a bearing on the work of archæologists and art-historians, this can only be constructive. Secondly, any attempt to do so would be misleading. Runic spelling is fluid and highly variable, particularly after the beginning of the Viking Age, when language change and the dramatic uptake of the Younger Fuþark meant that that a single rune could potentially represent a number of different vocalised sounds. More to the point, not enough is known about the pronunciation of Old English, Old Norse and related Germanic languages (not to mention regional accents and dialects) to make such phonetic representation in any way meaningful.