Daniel:Notebook/GenomeMiner/2013-8-28

From ZhangLabWiki
Jump to navigation Jump to search

HL155[edit]

Back to Calendar

Slim Indexing Analysis (Started 8/27/2013)[edit]

Counted Errors Using Modified Perl Script from Matt

Script: File:Imp count mismatch.txt

Raw Error Counts and Read Error Percentages

Total read counts and numbers of reads with errors

  Total reads Filtered reads Perfect match 1 ins/del 1 sub 2 ins/del 2 sub 1 ins/del & 1 sub 3+
Number 20924455 15399065 4801948 3829283 2242835 1891019 634644 1506743 492593
Percent NA 100 31.18 24.87 14.56 12.28 4.12 9.78 3.20

Error Percentages

Percent errors by base (ex 2% error means 1 in 50 bp will have an error)

  Mycroarray Slim Error %
Error Rate 2.03
Insertions 0.11
Deletions 1.18
Substitutions 0.75

Counting Results From Index 10-13[edit]

Print Only 3rd Column

awk '{print $3}' hrcp_sortfilt_slimindex_eq.sam > readlist.txt

Perl Script to Count Total Probe Number

perl script: File:Probecount.txt

perl probecount.plx > probecount_idx10to13.txt

Counts[edit]

Compare with full sequencing index

Histograms

Probe counts. Normalization for x axis is based on the expected number of reads, calculated by dividing the total number of reads for a given set by the total number of probes in that set. One indicates the probes were amplified more or less linearly, <1 means they were underamplified, and >1 means the probe was overamplifed. File:Probehist HL155 idx10to13.png

Pie Chart Statistics

Pie chart showing probe count statistics. Probe counts are divided into 3 classes: 0 reads, over-amplified (described as >5x the expected count), and normally amplified reads.

File:V6ProbeCounts Idx10to13.png|File:V4ProbeCounts Idx10to13.png

Discussion[edit]

At first I was a bit confused by these results, especially the pie charts. There are now more zeros and over-amplifications present, although there are fewer reads used. But I remembered I used the phred 64 flag, which is more stringent, so that probably threw out some data. Since the errors were probably only present one or a few times, they wouldn't have ended up in the zeros category (which had to be equal to zero) or in the over-amplification category (which required ~1000 reads), so they would have been in the normal read count category, which would have reduced the number of normal reads, decreasing normal read percent and increasing the other two.