2a:null 2b:["$","title",null,{"children":"Distinguishing 23andMe v3, v4, and v5 Chips by Genomic Position | lorenasandoval88 | Observable"}] 30:T295a,# Distinguishing 23andMe v3, v4, and v5 Chips by Genomic Position # Results v3 and v4 show substantial continuity, with 88.6% of v4 core coordinates also present in v3. In contrast, v5 represents a major shift in array content: 69.1% of its core coordinates are absent from both v3 and v4. Thus, the transition from v3 to v4 appears primarily as a reduction and refinement of an overlapping marker set, whereas the transition to v5 reflects a substantially different genomic coordinate design. ${await FileAttachment("image.png").image({width: 850})} ## Fingerprint Summary | Chip | Core markers | Shared with ≥1 other chip | Chip-specific | Shared across all 3 | |---|---:|---:|---:|---:| | **v3** | 959,702 | 607,163 | 352,539 | 105,026 | | **v4** | 596,247 | 536,264 | 59,983 | 105,026 | | **v5** | 622,753 | 192,201 | 430,552 | 105,026 | ## Composition of Each Chip Core | Chip | Shared | Chip-specific | |---|---:|---:| | **v3** | 63.3% | 36.7% | | **v4** | 89.9% | 10.1% | | **v5** | 30.9% | 69.1% | ### Shared vs. Chip-Specific Markers

v3 — 959,702 core markers

63.3% shared

36.7% specific

v4 — 596,247 core markers

89.9% shared

10.1% specific

v5 — 622,753 core markers

30.9% shared

69.1% specific

--- ## Chip-Specific Fingerprints These are the core genomic coordinates that distinguish each chip from **both** of the other chip versions. | Fingerprint | Unique coordinates | |---|---:| | **v3-specific** | 352,539 | | **v4-specific** | 59,983 | | **v5-specific** | 430,552 |

Number of chip-specific coordinates

v3	352,539
v4	59,983
v5	430,552

--- ## Pairwise Chip Overlap | Comparison | Shared coordinates | |---|---:| | **v3 ∩ v4** | 528,126 | | **v3 ∩ v5** | 184,063 | | **v4 ∩ v5** | 113,164 | | **v3 ∩ v4 ∩ v5** | 105,026 |

Pairwise overlap

v3 ↔ v4	528,126
v3 ↔ v5	184,063
v4 ↔ v5	113,164
All 3	105,026

### Understanding the Pairwise Counts The pairwise overlap values are **not mutually exclusive**. The **105,026 positions found on all three chips** are included within each pairwise overlap: - v3 ∩ v4 - v3 ∩ v5 - v4 ∩ v5 Therefore, the pairwise counts should not simply be added together. Subtracting the all-three intersection from each pairwise overlap gives the positions shared by **only two** chip versions: | Region | Coordinates | |---|---:| | **v3 only** | 352,539 | | **v4 only** | 59,983 | | **v5 only** | 430,552 | | **v3 + v4 only** | 423,100 | | **v3 + v5 only** | 79,037 | | **v4 + v5 only** | 8,138 | | **v3 + v4 + v5** | 105,026 | --- # Key Findings ## v3 The v3 core contains **959,702 genomic positions**, the largest of the three core manifests. Of these: - **607,163** are shared with v4 and/or v5. - **352,539** are specific to v3. - **36.7%** of the v3 core provides a v3-specific coordinate fingerprint. - v3 shares **528,126 positions with v4**, indicating substantial overlap between these two chip generations. ## v4 The v4 core contains **596,247 genomic positions**. Of these: - **536,264** are also present in v3 and/or v5. - **59,983** are specific to v4. - approximately **89.9%** of the v4 core overlaps at least one other chip. - approximately **88.6%** of the v4 core is also present in v3. v4 therefore has the **smallest chip-specific coordinate fingerprint** of the three versions. ## v5 The v5 core contains **622,753 genomic positions**. Of these: - **192,201** are shared with v3 and/or v4. - **430,552** are specific to v5. - approximately **69.1%** of the v5 core is absent from both v3 and v4. v5 therefore has the **largest chip-specific coordinate fingerprint** and is the most distinct of the three based on these core genomic positions. --- ## Overall Comparison

                         CORE MARKERS

v3    959,702
      ├── 607,163 shared
      └── 352,539 v3-specific

v4    596,247
      ├── 536,264 shared
      └──  59,983 v4-specific

v5    622,753
      ├── 192,201 shared
      └── 430,552 v5-specific

The overall pattern shows: - **v3 and v4 have substantial coordinate overlap.** - **v4 has relatively few positions unique to v4.** - **v5 is considerably more distinct from v3 and v4.** - **105,026 positions are represented in all three chip cores.** --- # Application to Unknown 23andMe Files The chip-specific coordinate sets can now be used as **reference fingerprints** for 23andMe files whose chip version is unknown. For an unknown genotype file, its genomic positions can be compared against:

v3-specific fingerprint: 352,539 possible positions
v4-specific fingerprint:  59,983 possible positions
v5-specific fingerprint: 430,552 possible positions

Because the fingerprint sets differ substantially in size, classification should use the **proportion of each fingerprint detected**, rather than comparing raw match counts alone. For example:

v3 match rate =
v3-specific coordinates found in unknown file
÷ 352,539

v4 match rate =
v4-specific coordinates found in unknown file
÷ 59,983

v5 match rate =
v5-specific coordinates found in unknown file
÷ 430,552

This allows chip version to be investigated using the **actual pattern of genomic positions contained in the file**, rather than relying only on filenames such as genome.txt. Unknown files should be compared using the same genome build because genomic coordinates can differ between genome assemblies. --- # Analysis Process ## Overview This analysis compared the **core genomic coordinates** from the 23andMe v3, v4, and v5 Build 37 chip manifests. Each marker is represented by its physical genomic position: `chromosome:position` For example: `1:752566` The comparison separated the core positions for each chip into two main groups: - **Shared markers** — positions also present in at least one of the other chip versions. - **Chip-specific markers** — positions present in that chip but absent from both other chip versions. The analysis also calculated: - pairwise overlap between v3 and v4 - pairwise overlap between v3 and v5 - pairwise overlap between v4 and v5 - positions shared across all three chips --- ## Input Core Manifests ```text gs://all23_0710/chip_manifests/v3_build37_95pct/coordinates_core.json gs://all23_0710/chip_manifests/v4_build37_95pct/coordinates_core.json gs://all23_0710/chip_manifests/v5_build37_95pct/coordinates_core.json ``` --- ## Workflow

v3 core coordinates ─────┐
                         │
v4 core coordinates ─────┼────► Compare chromosome:position
                         │
v5 core coordinates ─────┘
                               │
                               ▼
                    ┌─────────────────────┐
                    │ Coordinate overlap  │
                    └─────────────────────┘
                         │           │
                         ▼           ▼
                    Shared       Chip-specific
                    markers         markers
                         │           │
                         └─────┬─────┘
                               ▼
                    Chip fingerprint files

--- ## Run the Analysis The comparison script is saved as:

index.js

Run it with:

cd ~/chip-fingerprints
node index.js

The script: 1. reads the v3, v4, and v5 core coordinate manifests, 2. converts each coordinate list into a set, 3. compares the sets using `chromosome:position`, 4. identifies coordinates shared with at least one other chip version, 5. identifies coordinates specific to each chip, 6. calculates pairwise and three-way overlap, and 7. writes the resulting fingerprint files to Google Cloud Storage. --- ## Output Files

gs://all23_0710/chip_manifests/chip_fingerprints/
├── v3_chip_fingerprint.json
├── v4_chip_fingerprint.json
└── v5_chip_fingerprint.json

Each fingerprint contains:

chip core markers
│
├── sharedMarkers
│   └── also present in at least one other chip core
│
└── chipSpecificMarkers
    └── absent from both other chip cores

--- ## Generated Fingerprint Files

v3_chip_fingerprint.json
  ├── 607,163 shared markers
  └── 352,539 v3-specific markers

v4_chip_fingerprint.json
  ├── 536,264 shared markers
  └── 59,983 v4-specific markers

v5_chip_fingerprint.json
  ├── 192,201 shared markers
  └── 430,552 v5-specific markers

31:[{"id":"ed0649bb4e328c0cef0740a389b353a345f35b4c53205b80d02905bcd4383ec3760ac29b75723654a21a9f805e41bb95be7c5a8635b07f55ea49fe1aa0320793","href":"https://static.observableusercontent.com/files/ed0649bb4e328c0cef0740a389b353a345f35b4c53205b80d02905bcd4383ec3760ac29b75723654a21a9f805e41bb95be7c5a8635b07f55ea49fe1aa0320793","name":"index.js","type":"text/javascript","lastModified":1790699333279,"size":9896},{"id":"2879dfb2b3099c2bc42c2d31011cb25761d08ca24356cf01fd94f3283bf93be35de93c4ecdd087cf28b8c9119e7cc736a4750c5450ba6463b42b6797f0e162ab","href":"https://static.observableusercontent.com/files/2879dfb2b3099c2bc42c2d31011cb25761d08ca24356cf01fd94f3283bf93be35de93c4ecdd087cf28b8c9119e7cc736a4750c5450ba6463b42b6797f0e162ab","name":"image.png","type":"image/png","lastModified":1790700375283,"size":1193381}] 2e:{"id":"e06c47b44903d567","slug":"23and-me-v3-v4-and-v5-chip-fingerprints","title":"Distinguishing 23andMe v3, v4, and v5 Chips by Genomic Position","thumbnail":"ca08f84f7e1e64fd0931084e67e876481ead7d55a9610d9c34e0caeed2af5713","access_level":"public","create_time":"$D2026-09-28T14:14:07.177Z","update_time":"$D2026-09-29T17:46:31.270Z","view_time":null,"first_public_time":"$D2026-09-29T16:56:23.794Z","has_unpublished_edits":null,"fork":false,"stars":0,"starred":false,"owner":{"type":"team","id":"5a5147324ed15c62","login":"lorenasandoval88","name":"","avatar_url":"https://avatars-next.observableusercontent.com/c8926e2434569cc1f72aeac62929e824fdec3f0ec037d4a41a38a7728d14e950","tier":"starter_2024"},"version":68,"publish_version":68,"role":null,"authored":false,"authors":[{"type":"individual","id":"5c479a2c5ae5f52f","login":"lorenasandoval","name":"","avatar_url":"https://avatars-next.observableusercontent.com/c8926e2434569cc1f72aeac62929e824fdec3f0ec037d4a41a38a7728d14e950","tier":null}],"members":[{"type":"team","id":"5a5147324ed15c62","login":"lorenasandoval88","name":"","avatar_url":"https://avatars-next.observableusercontent.com/c8926e2434569cc1f72aeac62929e824fdec3f0ec037d4a41a38a7728d14e950","tier":"starter_2024","role":"editor"}],"comments":[],"custom_thumbnail":false,"trashed":false,"lapsed":null,"subscribed":null,"forks":0,"collection":{"id":"1476de78fbd46e52","slug":"23andme-data","title":"23andme-open-data","owner":{"id":"5a5147324ed15c62","login":"lorenasandoval88","name":""}},"type":"notebook","body":{"titling":"auto","title":"Distinguishing 23andMe v3, v4, and v5 Chips by Genomic Position","cells":[{"id":3,"value":"$30","pinned":true,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","output":"$undefined","database":"$undefined"},{"id":11,"value":"","pinned":true,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","output":"$undefined","database":"$undefined"},{"id":7,"value":"","pinned":true,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W31","stdlib":"2"}} 2f:"$2e:body"