Sreenivas Eadara
Human-readable sequence formats are required to allow users to view the results of ultratherm.
However, ultratherm (and perhaps other programs) may require additional data associated with each nucleotide of each sequence, which are required in order for a human-readable sequence format to be used as an input file. This specification is designed to utilize the FASTQ standard format to store sequences and 5 bits of data per nucleotide. The protocol may be expanded in the future.
The FASTQ format allows phred_quality scores, represented as a string of ASCII characters between ! and ~, to be stored alongside nucleotide sequences, such that each nucleotide has an associated quality (from 0 to 93 inclusive).
In this specification, only a subset of the FASTQ characters (characters between ! and ~) are to be used. Only alphanumeric characters are permitted. 5 consecutive bits are encoded per character in b1, b2, b3, b4, b5. The sequence of bits is as follows:
[10XXXXX]
Where XXXXX are the 5 least significant bits.
In ultratherm, this will correspond to the following (0: false, 1: true)
(note that ASCII is 7-bit. When stored in a byte the MSB is always 0).
| b5 | b4 | b3 | b2 | b1 |
|---|---|---|---|---|
| 0 (False) | 0 (False) | 0 (False) | 0 (False) | 0 (False) |
| 1 (True) | 1 (True) | 1 (True) | 1 (True) | 1 (True) |
| score_this_nucl | is_rna | score_region | no_indel | no_mod |
It is possible to map each byte of data (i.e. each ASCII character) to a PHRED score. The ASCII ‘@’, which is the first character in this specification, corresponds to ASCII 64 and PHRED score 31. In order to obtain a PHRED score from a list of bits b1-b5, one just needs to convert the list of bits to a 5-bit integer and add 31, obtaining the PHRED score.
This specification supports the addition of a specific gap character such that multiple sequences (i.e. hybrids) can be stored within the same record. This is accomplished with the ‘&’ character as follows:
@16 0.9314516162927949
GAAGCAGACAAGAGCAGUCCUAGUCCUGAUUGCCAAAAGGCGAAUUAUUA&GGACUAGCUCUU
+
___XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX&LLLLLLLLLLLL
This can be accomplished only because & is a permissible PHRED score (5) and such characters are permitted in BioPython as well as other parsers. The ‘&’ character was selected because ViennaRNA permits multiple strand entry using the ‘&’ character and this increases the convenience for manual review of ultratherm sequences.