search

twolinin / longphase

GNU General Public License v3.0
98 stars 6 forks source link

Pochung #79

Closed benson10120 closed 1 week ago

benson10120 commented 2 weeks ago

Update ModCall and ReadCorrection tagging mechanism

benson10120 commented 1 week ago

In the program ModCallParsingBam.cpp, within the function exportResult, modify the selection mechanism of ModCall so that methylation site are selected only if both consecutive positions on CpG island meet the criteria with another CpG island.

In the program PhasingGraph.cpp, within the function findBestEdgePair, if the situation involves a methylation site connecting to an SNP or an SNP connecting to a methylation site, change the edgeThreshold from 0.7 to 0.3.

In the program PhasingGraph.cpp, within the function ReadCorrection, adjust the weights of SNP, SV, MOD, and INDEL in ReadCorrection. SNP and SV has weight 1, INDEL has weight 0.1 and MOD has weight 0.

Before modification, switch error values for 10x-60x are 1123, 1286, 1262, 1279, 1287, 1292, respectively. After modification, switch error values for 10x-60x are 1124, 1264, 1233, 1255, 1262, 1272, respectively. Before modification, n50 values for 10x-60x are 808264, 1814440, 2658491, 3384448, 3969440, 5329227, respectively. After modification, n50 values for 10x-60x are 809291, 1814734, 2677820, 3371590, 3920477, 5348064, respectively.

twolinin commented 1 week ago
  1. Explain the addition of the overlapThreshold parameter in phasing.
  2. The method for determining overlap in addEdge seems changed. If so, please update the document.
  3. Please remove empty file modphase-30x-test0703-1_info
  4. Please explain the types of variants and weighted using comments https://github.com/twolinin/longphase/pull/79/files#diff-1d209d89291d59cc5f6680c8c0b56cefb3cf5ac76a3875455337e995926bf031R609