Skip to content

Commit a7bae5f

Browse files
committed
concept-live-call audit, one small content fix. the only stale claim was the example output block — used legacy V allele names IGHV3-23*04 / IGHV3-23*01 that don't exist in the current shipped configs (verified absent from refdata) — replaced with names that are real in the current scheme (IGHVF10-G38*04 for the single-confident case, IGHVF6-G24*03 / IGHVF6-G24*04 for the same-gene paralog tie) sourced from an actual seed=42 mutate(count=25)+corrupt_pcr(count=8) run. also corrected one slip: the "honest drift" example was labelled v_call but the values shown were D alleles — now correctly labelled d_call. the D examples in the drift block (IGHD6-13*01, IGHD6-19*01, IGHD6-25*01) and the part-04 paralog example (IGHD4-11 / IGHD4-4 sibling genes) were verified to be real D alleles in the current shipped config so they stay. the rest of the page held up: the score-and-tie semantics, the structural+elastic walk, the two walker caps (trim cap + structural-boundary cap), and the truth-first projection via projected_allele_id + live_call_name are all accurate against the engine code i worked on for the Y6 caller. snippet's for rec in records: iteration over SimulationResult verified to work — records iterate as dicts directly. count=int form of .mutate() and .corrupt_pcr() also verified.
1 parent de87d7a commit a7bae5f

1 file changed

Lines changed: 5 additions & 4 deletions

File tree

website/concept-live-call.html

Lines changed: 5 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -113,14 +113,15 @@ <h2>Why GenAIRR records two call fields per segment.</h2>
113113
<div class="ph-card ph-card-code">
114114
<div class="ph-card-head">What you see in the record now</div>
115115
<pre class="code"><code><span class="c-c"># single confident call</span>
116-
v_call <span class="c-s">"IGHV3-23*04"</span>
116+
v_call <span class="c-s">"IGHVF10-G38*04"</span>
117+
truth_v_call <span class="c-s">"IGHVF10-G38*04"</span>
117118

118119
<span class="c-c"># tie-set — paralogs the surviving bases can't tell apart</span>
119-
v_call <span class="c-s">"IGHV3-23*04, IGHV3-23*01"</span>
120-
truth_v_call <span class="c-s">"IGHV3-23*01"</span>
120+
v_call <span class="c-s">"IGHVF6-G24*03, IGHVF6-G24*04"</span>
121+
truth_v_call <span class="c-s">"IGHVF6-G24*03"</span>
121122

122123
<span class="c-c"># honest drift — truth dropped out of the evidence tie</span>
123-
v_call <span class="c-s">"IGHD6-13*01, IGHD6-19*01"</span>
124+
d_call <span class="c-s">"IGHD6-13*01, IGHD6-19*01"</span>
124125
truth_d_call <span class="c-s">"IGHD6-25*01"</span></code></pre>
125126
</div>
126127
</div>

0 commit comments

Comments
 (0)