Glossary
- BLAST
- Basic Local Alignment Search Tool, used here only to nominate RNA for exact evaluation.
- CPM
- Counts per million, an abundance normalized within one sample.
- FASTA
- A sequence text format; collapsed FASTA records encode an RNA sequence and its raw count.
- HSP
- High-scoring segment pair, a local alignment reported by BLAST and retained as provenance.
- IUPAC
- The standard nucleotide symbols, including letters that represent more than one possible base.
- nt
- Nucleotide, the unit used for probe and RNA length.
- PAGE
- Polyacrylamide gel electrophoresis; the on-screen blot is not a physical PAGE calibration.
1. Input, normalization, and abundance
Each sample is a collapsed FASTA library. RNA is converted to uppercase, input T is changed to U, and identical normalized sequences are merged by adding their raw counts. Standard IUPAC symbols are valid and remain in the library; any other sequence character is a validation error.
Ambiguity is retained
A valid IUPAC symbol never removes an RNA by itself. In particular, N is neither an automatic match nor a whole-read mismatch.
The probe is an unambiguous DNA sequence written 5′→3′, containing only A, C, G, and T, with a length of 10–35 nt.
| Parameter | Value | Effect |
|---|---|---|
| Minimum raw count | 10 by default | RNA below the selected cutoff is removed before abundance calculation. |
| RNA length | 10–150 nt selectable; 10–150 nt by default | Sets the accepted read range and freezes the displayed length axis. |
| DNA probe length | 10–35 nt | Sets the full target length; a nominated RNA may be up to 3 nt shorter when at least 80% remains as one terminal overlap. |
After the raw-count cutoff, the post-cutoff sample total is the sum of all retained RNA counts. Read-length policy and IUPAC symbols do not alter that denominator. Each retained RNA is normalized as:
The expected RNA target is the reverse complement of the submitted DNA probe, written 5′→3′. For candidate search only, both query and subjects use a technical nucleotide representation in which U becomes T; IUPAC ambiguity is preserved. Final qualification returns to RNA notation.
2. BLAST candidate nomination
Local BLAST+ 2.17.0+ nominates RNA that remains after the raw-count and selected RNA-length filters. The post-cutoff sample total used for CPM still includes retained-count RNA outside the selected length range. Candidates are deduplicated at RNA level, while the retained HSP records preserve the local-alignment evidence for audit.
| Short-query task | blastn-short |
|---|---|
| Word length | 4 |
| Match / mismatch score | 1 / −3 |
| Gap-open / gap-extension cost | 5 / 2 |
| E-value | 1000 |
| Masking | Disabled |
| Orientation | Plus strand |
| Reporting | One HSP per query–subject pair; no fixed top-N truncation |
Nomination is not qualification
BLAST is a heuristic prefilter. Its scores, identity, coverage, and E-value do not decide the final match. No BLAST candidate is not proof that the RNA has more than six mismatches.
3. IUPAC-aware Hamming qualification
For nominated RNA at least as long as the probe, every complete probe-length window is compared position by position with the expected RNA target. A shorter nominated RNA is also evaluated when no more than 3 terminal nucleotides are missing and integer coverage is at least 80%; its complete sequence is compared with every same-length slice of the expected target. Overlaps are contiguous and ungapped, so internal deletions receive no gapped rescue.
IUPAC symbols and possible bases
| Symbol | Possible RNA bases |
|---|---|
A | A |
C | C |
G | G |
U | U |
R | A, G |
Y | C, U |
S | C, G |
W | A, U |
K | G, U |
M | A, C |
B | C, G, U |
D | A, G, U |
H | A, C, U |
V | A, C, G |
N | A, C, G, U |
At each position, the minimum cost is zero when the expected base is among the symbol’s possible bases, otherwise one. The maximum cost is zero only when the symbol identifies exactly the expected base, otherwise one. Summing across the window gives its minimum and maximum possible mismatch counts.
The selected overlap has, in order, the lowest maximum mismatch count, the lowest minimum mismatch count, the earliest expected-target overlap, and the leftmost RNA start. Missing terminal nucleotides are reported separately and excluded from Hamming. Ambiguity outside the selected overlap does not affect its result. These bounds describe logical possibilities, not probabilities.
How N is evaluated
Inside a selected overlap, N includes the expected base among its possibilities but also allows alternatives, so it can widen the mismatch bounds. Outside that overlap, it has no effect on those bounds.
Cumulative mismatch thresholds
The available thresholds are 0, 1, 2, 3, 4, 5, 6. At a chosen threshold:
- Verified
- The maximum possible mismatch count is at or below the threshold. The RNA contributes its CPM once.
- Possible
- The minimum is at or below the threshold but the maximum is above it. The RNA remains diagnostic and contributes no blot signal.
- Rejected
- The minimum possible mismatch count is above the threshold.
Thresholds are cumulative: increasing the threshold retains previously verified RNA. Multiple HSPs or equally good windows never multiply an RNA’s contribution.
4. Bands, axis, and display scaling
Verified, signal-eligible RNA is grouped into exact-length bands. The submitted RNA-length range is frozen as the logarithmic axis for the analysis, including empty views. Equal submitted bounds expand by ±5 nt. Markers use 5 nt intervals for spans up to 50 nt and 10 nt intervals for wider spans.
This is an illustrative migration axis, not a physical PAGE calibration.
Presentation-only scaling
Display scaling changes pixel brightness only. It does not change RNA classification, CPM, bands, or exported scientific values.
| Choice | Reference rule | Current values |
|---|---|---|
| Fixed source scale | The same source-signal mapping is used for compatible analyses. | Background 0.05; factor 0.0005; display cap 0.7; source clipping threshold 1300.0. |
| Automatic exposure | Uses positive band peaks from threshold 6 across all lanes. Fewer than 20 peaks use the maximum; 20 or more use the interpolated 95th percentile. No peaks produce no signal factor. | Background 0.05; target 0.65; headroom 1.5×; display cap 0.95. |
5. Results, provenance, and boundaries
Result views provide the illustrative blot, exact-length bands, verified and diagnostic RNA tables, selected-overlap details, and retained BLAST HSP provenance. Images can be downloaded as SVG and PNG. The portable .isnb archive carries the parameter snapshot, nominated RNA result tables, provenance, visual data, and checksums needed for later inspection. Keep the submitted FASTA libraries separately; the archive does not contain them.
The local .isnb viewer reads that saved archive in the browser. It neither uploads the archive nor reruns BLAST or mismatch evaluation.
Interpretation limits
The model covers full-length substitution comparisons after heuristic candidate nomination. It does not model FASTQ-derived base probabilities, insertions or deletions, duplex structure, melting temperature (Tm), Gibbs free energy (ΔG), probe chemistry, wet-lab hybridization conditions, or physical PAGE calibration. Predictions require experimental validation.