Ylaine/2009-8-26
Jump to navigation
Jump to search
Clustering[edit]
- Re-ran clustering with similar results. Only difference is 1 false SNP that was accepted in the first round instead of the second round.
File:Aug17 nodbSNP.jpg File:Aug17 cluster1.jpg File:Aug17 cluster2.jpg
- Commands:
[error, groups] = snpCluster(match,miss);
- 328 accepted, 3102 rejected, total error rate of 8.45% (46.46% MD, 1.42% FP)
- 287 true and 41 false
groups_match = groups(1:536); rejects_match = match(find(groups_match==2),:); groups_miss = groups(537:3430); rejects_miss = miss(find(groups_miss==2),:); [error, groups2] = snpCluster(rejects_match, rejects_miss);
- 214 accepted, 2888 rejected, total error rate of 6.61% (48.19% MD, 2.98% FP)
- 129 true and 85 false
Testing Filter[edit]
- Validate method on another test set:
/Users/ajgore/YlaineReads/s_5_sequence.bowtie.sam.variants.txt grep -v 'rs' orig.snp > nodbsnp ../Scripts/compare1K.pl nodbsnp > 1k
- Remove extraneous information:
grep -v '#' 1k>1kcompare
- Partition based on matching (lump mismatches with misses
grep ' match' 1kcompare > nodbsnp.match grep -v ' match' 1Kkcompare > nodbsnp.miss
- Out of 21322, only 66 match 1000 Genomes: clustering will probably not be useful
False Negatives[edit]
cat gold.covereda*>gold.covered ../Scripts/matchGold.pl gold.covered>gold.match
- 9553 total: 7869 match, 1182 miss, 512 mismatch
../Scripts/threshold.pl gold.match 2 > gold.match2
' | Correct SNP | ' | Incorrect | ' | Missed | ' | Total |
1x | 7859 | 82.27% | 512 | 5.36% | 1182 | 12.37% | 9553 |
2x | 7715 | 85.97% | 348 | 3.88% | 911 | 10.15% | 8974 |
8x | 6732 | 94.05% | 57 | 0.80% | 369 | 5.16% | 7158 |
20x | 4997 | 96.37% | 10 | 0.19% | 178 | 3.43% | 5185 |
Poster[edit]
- Determine whom to acknowledge for experimental results