# The bigBed columns `cas12a.bb` is `bed9+14` = **23 fields**, one row per guide, defined by `cas12aTargets.as`. A real row (hg38, `chr1:1000166`): ```text chrom chr1 chromStart 1000166 0-based start of the 27-nt PAM+spacer site chromEnd 1000193 chromStart + 27 name (blank) score 96 drives the browser score filter (see below) strand + thickStart 1000170 spacer start (chromStart + 4); the thin 4 nt is the PAM thickEnd 1000193 itemRgb 0,200,0 display color (see below) guideSeq CCACACTCGCCCCAGCCAATCGA 23-nt spacer pam TTTC 4-nt PAM unique_TTTV True no identical protospacer elsewhere at a TTTV PAM unique_TTTN True ... at a TTTN PAM TTTV_enCas12a_specificity 90% (0.9377) TTTN_enCas12a_specificity 85% (0.7001) TTTV_AsCas12a_specificity 89% (0.9958) TTTN_AsCas12a_specificity 87% (0.9628) ascas12a_deepcpf1_score 96% (0.8212) enseq_deepcpf1_score 61% (0.6454) deepcpf1_score 57% (53.5321) enpam_gb_score 19% (0.7867) _mouseOver EnCas12a Spec: 85, AsCas12a Spec: 87, EnCas12a-DeepCpf1: 61, ... _offset 27614949 byte offset into crisprDetails.tab (0 = no list) ``` ## The eight score columns read `percentile (raw)` - **percentile** ranked within the guide's chunk. - **raw** is exact and global: specificity on 0-1 (`1.0` = no off-targets); DeepCpf1 is the exception, its native output is ~0-100. Four **off-target specificity** columns = {enCas12a, AsCas12a} x {scanned vs the TTTV, TTTN index}. Four **on-target efficiency** columns = AsCas12a-DeepCpf1, EnCas12a-DeepCpf1, DeepCpf1, EnPAM-GB. ## score and color - **score** (field 5, 0-1000): `round(AsCas12a TTTN specificity_raw x 100)`, e.g. `0.9628 -> 96`. The browser's score filter is therefore a specificity filter. - **itemRgb** (field 9): ramps on the **AsCas12a-DeepCpf1 percentile**, green `>75`, yellow `>50`, red `<=50`; overridden to grey for low specificity, then 1/2-mismatch, then non-unique. The row above is green because AsCas12a-DeepCpf1 = 96%. ```{warning} Percentiles are per-chunk (see the [scoring step](../ucsc_cas12a/03_score_track.md)), so `score`, `itemRgb`, and the `NN%` are ranked within a chunk, not genome-wide. The parenthesized **raw** values are exact and comparable across the whole genome. ```