Skip to content

Commit

Permalink
Add truth version to documentation.
Browse files Browse the repository at this point in the history
PiperOrigin-RevId: 329941821
  • Loading branch information
pichuan authored and copybara-github committed Sep 3, 2020
1 parent 0f3a999 commit d6caaef
Showing 1 changed file with 7 additions and 5 deletions.
12 changes: 7 additions & 5 deletions docs/metrics.md
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@ total | 431m = 6.7 hours

### Accuracy

hap.py results on HG002 (all chromosomes).
hap.py results on HG002 (all chromosomes), using NIST v4.2 truth.

Type | # TP | # FN | # FP | Recall | Precision | F1_Score
----- | ------- | ----- | ---- | -------- | --------- | --------
Expand All @@ -38,7 +38,7 @@ total | 21m

### Accuracy

hap.py results on HG002 (all chromosomes).
hap.py results on HG002 (all chromosomes), using NIST v4.1 truth.

Type | # TP | # FN | # FP | Recall | Precision | F1_Score
----- | ------- | ---- | ---- | -------- | --------- | --------
Expand All @@ -48,6 +48,8 @@ SNP | 38180 | 396 | 116 | 0.989735 | 0.996975 | 0.993341

## PacBio (HiFi)

### Runtime

Runtime is on HG002 (all chromosomes).

Stage | Time (minutes)
Expand All @@ -59,7 +61,7 @@ total | 421 = 7 hours

### Accuracy

hap.py results on HG002 (chr20).
hap.py results on HG002 (chr20), using NIST v4.2 truth.

(The input BAM is haplotagged already and DeepVariant
was run with `--sort_by_haplotypes=true --parse_sam_aux_fields=true`.)
Expand All @@ -84,8 +86,8 @@ total | 455 m = 7.6 hours

### Accuracy

Evaluating on HG003 (all chromosomes), which was held out while training the
hybrid model.
Evaluating on HG003 (all chromosomes, using NIST v4.2 truth), which was held out
while training the hybrid model.

Type | # TP | # FN | # FP | Recall | Precision | F1_Score
----- | ------- | ---- | ---- | -------- | --------- | --------
Expand Down

0 comments on commit d6caaef

Please sign in to comment.