Daniel:Notebook/GenomeMiner/2013-9-17: Difference between revisions

From ZhangLabWiki
Jump to navigation Jump to search
>Djacobse
>Djacobse
 
(4 intermediate revisions by the same user not shown)
Line 34: Line 34:
==Errors Bases 30 to 50==
==Errors Bases 30 to 50==


Rewrote part of the fastq error generating script to choose ranges of bases to have wrong.  Can now insert errors wherever we please at any percentage we want.  I selected the error to be the same as the mimic data.
Rewrote part of the fastq error generating script to choose ranges of bases to have wrong.  Can now insert errors wherever we please at any percentage we want.  I selected the error to be the same percentages as the mimic data, but only from bases 30 to 50.


===Workflow===  
===Workflow===  
Line 42: Line 42:
  3. hl155bash.sh
  3. hl155bash.sh
  4. perl imp_count_mismatch.plx (Matt's error counting script)
  4. perl imp_count_mismatch.plx (Matt's error counting script)
'''MATLAB error counts''':
628508 substitutions (0.42 pct)
222963 insertions (0.15 pct)
42735 deletions (0.03 pct)
The errors were meant to be mimic percent (1.01%s, 0.26%i, 0.05%d), but the actual percentages are lower, since the percentages given are probabilities, and automatcally bases 1:29 were without error.
===Alignment Results===
2959000 reads; of these:
  2959000 (100.00%) were unpaired; of these:
    34339 (1.16%) aligned 0 times
    2919032 (98.65%) aligned exactly 1 time
    5629 (0.19%) aligned >1 times
98.84% overall alignment rate
===Error Counting Results===
Error Rate: 0.532%
Error Rate of Insertions: 0.129%
Error Rate of Deletions: 2.98e-04%
Error Rate of Substitutions: 0.403%
Looking at the basic error counting result, and comparing it to the MATLAB error counts before, the results are actually very close.  Once again deletions has been undercounted by about 2 orders of magnitude, but substitutions and insertions are close, off by only about 0.02% each.

Latest revision as of 00:05, 18 September 2013

Mock HL155 (Started 9/9/2013)[edit]

Back to Calendar

High Deletion Rate Test[edit]

Checking the results from the previous run by redoing the data, this time using a lower substitution rate and a higher deletion rate. Substitutions: 0.05%, Insertions: 0.26%, Deletions, 1.01%.

Workflow[edit]

1. MockHL155_Master.m, Switch 2
2. scp v4s1_v4s1_mockseq_error_losubhidel.fq djacobse@132.239.135.41:/media/LTS_15T/DEJ_LTS/SeqStore/130628_HL155/mockseq/hidellowsub/
3. hl155bash.sh
4. perl imp_count_mismatch.plx (Matt's error counting script)

Alignment Results[edit]

2959000 reads; of these:
 2959000 (100.00%) were unpaired; of these:
   523061 (17.68%) aligned 0 times
   2431720 (82.18%) aligned exactly 1 time
   4219 (0.14%) aligned >1 times
82.32% overall alignment rate

So much poorer alignment than high substitutions, but still overall a high rate.

Error Counting Results[edit]

Error Rate: 0.523%
Error Rate of Insertions: 0.252%
Error Rate of Deletions: 0.016%
Error Rate of Substitutions: 0.255%

Errors Bases 30 to 50[edit]

Rewrote part of the fastq error generating script to choose ranges of bases to have wrong. Can now insert errors wherever we please at any percentage we want. I selected the error to be the same percentages as the mimic data, but only from bases 30 to 50.

Workflow[edit]

1. MockHL155_Master.m, Switch 5
2. scp  djacobse@132.239.135.41:/media/LTS_15T/DEJ_LTS/SeqStore/130628_HL155/mockseq//
3. hl155bash.sh
4. perl imp_count_mismatch.plx (Matt's error counting script)

MATLAB error counts:

628508 substitutions (0.42 pct) 
222963 insertions (0.15 pct) 
42735 deletions (0.03 pct)

The errors were meant to be mimic percent (1.01%s, 0.26%i, 0.05%d), but the actual percentages are lower, since the percentages given are probabilities, and automatcally bases 1:29 were without error.

Alignment Results[edit]

2959000 reads; of these:
 2959000 (100.00%) were unpaired; of these:
   34339 (1.16%) aligned 0 times
   2919032 (98.65%) aligned exactly 1 time
   5629 (0.19%) aligned >1 times
98.84% overall alignment rate

Error Counting Results[edit]

Error Rate: 0.532%
Error Rate of Insertions: 0.129%
Error Rate of Deletions: 2.98e-04%
Error Rate of Substitutions: 0.403%

Looking at the basic error counting result, and comparing it to the MATLAB error counts before, the results are actually very close. Once again deletions has been undercounted by about 2 orders of magnitude, but substitutions and insertions are close, off by only about 0.02% each.