From Genome Analysis Wiki
Jump to navigationJump to search
1,316 bytes added
, 16:10, 23 February 2012
Line 12: |
Line 12: |
| * Only works on one chromosome at a time | | * Only works on one chromosome at a time |
| * Repeats are single-base only | | * Repeats are single-base only |
| + | ** An unrepeated base has repeat count = 0 ('A') |
| + | ** A base repeated once has repeat count = 1 ('AA') |
| * Skips reference positions with base 'N' | | * Skips reference positions with base 'N' |
| | | |
Line 20: |
Line 22: |
| A position is considered to have an Insertion Discordance if at least 1 read has an insertion following this position AND at least 1 read does not AND there are at least the "minimum depth" reads at this position. | | A position is considered to have an Insertion Discordance if at least 1 read has an insertion following this position AND at least 1 read does not AND there are at least the "minimum depth" reads at this position. |
| | | |
− | == | + | == Error Rate Algorithm == |
| + | |
| + | The weighted average error rate, weighted average deletion error rate, and weighted average insertion error rate are calculated for each repeat count. |
| + | |
| + | The error rate is weighted by the depth, so the discordant counts are |
| + | |
| + | |
| + | For each Repeat Count in the File |
| + | For each Depth at this Repeat Count in the file |
| + | if (depth > MaxAllowedDepth) |
| + | skip calculating error rate for this depth |
| + | else |
| + | // Note: numDiscordant is not equivalent to numDeleteDiscordant + numInsertDiscordant since a position could have both types of discordants |
| + | numDiscordant = number of discordant positions with this repeat count and depth |
| + | numDeleteDiscordant = number of deletion discordant positions with this repeat count and depth |
| + | numInsertDiscordant = number of insertion discordant positions with this repeat count and depth |
| + | count = number of positions with this repeat count and depth |
| + | |
| + | <math>errorRate = 1 - (\tfrac{numDiscordant}{count})^{({\tfrac{1}{depth}})}</math> |
| + | sumErrorRates += errorRate * count * (depth-1) |
| + | numErrorRates += count * (depth-1) |
| + | Repeat Count Weighted Error Rate = sumErrorRates/numErrorRates |
| + | |
| | | |
| = Usage = | | = Usage = |