User:Expresso/Cpg220kProbeSetExpansion
Additional Probes to Cpg220k[edit]
Probe Script Issues[edit]
(1) The cutoffs for CpGs within targeting arms was too relaxed.
• I updated the script so that a strict CpG cutoff is applied for CpGs present across both arms. This cutoff is now an input variable.
(2) The script produces overlapping probes.
• This problem is less of an issue. The IsCompatibleProbes() subroutine checked for overlap between targeting arms (strand insensitive) and overlap between target regions (strand sensitive). However, it did not check for overlap between targeting arms of one candidate probe and the target region of another probe. This led to the output of overlapping probes.
(3) Bisulfite conversion misses a small number of CGs. After in silico bisulfite conversion of the sequence, all Cs within CpGs remain C while all other cytosines are converted to Ts. However, there are a small number of cytosines present in CGs sites that are converted thymine. I did not fix this issue. Since I only added probe regions that contained no CpGs in the targeting arms, I just excluded the small number of probes whose arm sequences contained no Cs but whose arms covered CpG sites.
New Changes in Probe Generation[edit]
Two regions, LowEfficiency and RefExpansion, are used for probe generation three times. Each time, the maximum CpG cutoff in the targeting arms is adjusted. These values range from 0-2. A value of 1 means that only one CpG is tolerated across both arms.
I plan to integrate new regions produced by the probe generation iteration that allowed no CpGs in targeting arms first. I will check the number of probes included and the increase in coverage. Then I will include non-overlapping regions that contain at most 1 CpG and finally, regions that include 2 CpGs across both targeting arms.
• This will allow for a maximal increase in coverage with a minimal increase in probe production.
Procedure (2010-01-04)[edit]
(1)
• Using the DMR220kRegions.bed file, the probe output files, and the low efficiency probes (s5Cpg220k, s6Cpg220k, and s7Cpg220k files), I found the regions I needed to resubmit for probe generation.
• Using the refFlat.txt file for hg18, I a generated a list of 4kb regions [-2kb from TSS, 2kb from TSS] for all unique TSSs (script).
• Using the I downloaded the snomiRNA File from UCSC (hg18), I generated a list of all regions that spanned 2 kbp around the miRNA/snoRNA regions (miRNA/snoRNA Candidate Regions). The file contains 1120 different miRNAs and snoRNAs.
(2)
• The probe generation script was modified to read in the sequences for the low efficiency targeting arms. I modified the generation script so that generated probes that contained any of those low efficiency sequences were ignored.
• I found all CpGs and SNPs located within RefSeq Expansion (CpG script and SNP script). Probe generation script for this set did not filter for "low efficiency" targeting arm sequences.
• Probe generation for miRNA is analogous to the RefSeq Expansion method. miRNA/snoRNA Covered CpGs and SNPs
• All three sets were submitted for three rounds of probe generation. The rounds allowed for 0, 1, and 2 total CpGs in the targeting arm sequences.
Process Input Files For Probe Generation Script
(3)
• Generated probes were checked for same strand overlapping issues (script). If two probes overlapped, the probe with: (1) more CpGs in targeted arms, (2) lower CpG coverage, and (3) lower SNP coverage was discarded.
All Generated Probes (0, 1, 2 CpGs on Targeting Arms)
Non-Overlapping Generated Probes
(4)
• All targeting arms from non-overlapping probes produced in (3) were mapped back to the genome via SOAP. All mapped locations were reported. If a targeting arm mapped to more than one location, the probe was outputted to a separate file.
Convert Probe Sequence Files to FASTA Script
Find Multiple Mapped Probes Script
Uniquely Mapping Non Overlapping Probes
• Probes that contained non-uniquely mapping targeting arms were filtered out. 7.4% of probes (47,629 out of 645,027 probes total) contained at least one non-uniquely mapping targeting arm.
Results (2010-01-04)[edit]
The original DMR220kRegions.bed file contains 119,473 regions.
51,409 regions are in the s5, s6, and s7 sets (s5 contains 30,789 regions, s6 contains 13,351 regions and s7 contains 7,211 regions).
• 68,064 regions remain.
Expanded Cpg220k Probe Set Includes Probes That Were Added in the Following Order:
• 37,793 probes that target Low Efficiency regions with 0 CpGs located on targeting arms (37,793 regions).
• 66,093 probes that target RefSeq gene regions with 0 CpGs located on targeting arms (66,093 regions);
• 34,154 probes that target RefSeq gene regions with at most 1 CpG located on either target arm and each target regions contains at least 5 CpGs (18,295 regions).
• 16,083 probes that target Low Efficiency regions with at most 1 CpG located on either target arm and each target region contains at least 3 CpGs (8,748 regions).
• 1,794 probes that target miRNA/snoRNA regions with 0 CpGs across targeting arms (1,794 regions).
• 1,921 probes that target miRNA/snoRNA regions with at most 1 CpG across targeting arms (1,343 regions).
• 7,154 probes targeting RefSeq gene regions with at most 2 CpGs across targeting arms (4,180 regions).
• 8 probes targeting ALDH3A2 (2 regions).
Script To Add 2 CpG RefSeq Probes
Script To Attach RefSeq Genes to BED Regions
Script to Merge Individual Probe Files
• 165,000 Probes Total.
• An additional 138,248 regions covered in the expanded probe set (206,312 regions total excluding low efficiency probes from the original 220k probe set).
• 32,456,350 bp covered by the old and new Cpg220k probes.
• 976,737 CpGs covered by the old and new Cpg220k probes.
• 189,697 dbSNP130 candidate SNP sites covered.
• 183,537 CpGs covered on both strands.
• 28,596 RefSeq IDs covered out of 30,302 (94.5%).
Probe Sequence File for Probe Generation
Probe Sequence File with Annotation
BED File to Visualize Targeted Locations
BED File to Visualize Only the Cpg220k Expansion Targeted Locations (138,248 regions)
Checks (2010-01-04)[edit]
• New probes do not match low efficiency probes (Script)
• New probes do not overlap with each other (Script)