Chris:LabNotes/FateMapping/Calendar/2015/2015-7-2
Jump to navigation
Jump to search
Verify filtered 20140922 probes[edit]
Background[edit]
- Previously, we decided on filtering the probes designed on 20140922. This was done by taking the top 2000 probes from either gDNA, gDNA MDA, or C1 single cell capture then removing the very top 20. This is described on <http://genome-tech.ucsd.edu/LabNotes/index.php/Chris:LabNotes/FateMapping/Calendar/2015/2015-5-23>. Theoretically, by cherry picking the top probes and removing the very top, we should get much better probe capture/more even bias compared to the larger probe set.
- However, upon probe capture on new BA17 C1 single cell amplicons done on <http://genome-tech.ucsd.edu/LabNotes/index.php/Chris:LabNotes/FateMapping/Calendar/2015/2015-6-15>, we did some analysis on the sequencing data and it looks like bias actually turned out to be worse.
- We will be verifying probe design today. One potential concern is that the probes may have been designed/ordered differently than originally done, so we will be looking at the scripts that Dr. Zhang used to design these probes and also the final ordering file sent to CustomArray
- As a reminder, the two different probe sets were used on different single cell amplicons from BA17 (2015_02_26 C1 run):
Well # Probe Set Capture Link G1 Original http://genome-tech.ucsd.edu/LabNotes/index.php/Chris:LabNotes/FateMapping/Calendar/2015/2015-3-13 (Strip A = Hemo, Strip B = Phusion) H1 "" "" A2 "" "" B2 "" "" C2 "" "" D2 Filtered http://genome-tech.ucsd.edu/LabNotes/index.php/Chris:LabNotes/FateMapping/Calendar/2015/2015-6-15 E2 "" "" F2 "" "" G2 "" "" A3 "" ""
Verify Scripts[edit]
- Below are the scripts used when Dr. Zhang finalized the filtered probe sets for order at Custom Array:
File:Cw 20150702 Probe2padlock13May2014V4.txt 20150523 New Probes (new set of 8000 not used in this analysis, v4) File:Cw 20150702 Probe2padlock13May2014V6.txt Filtered Hemo Probes (v6) File:Cw 20150702 Probe2padlock13May2014V8.txt Filtered Phusion Probes (v8)
- Below is the script used to make the original probe order from 20140922:
File:Probe2padlock30Sept2014V6.txt
- There are a few differences between the new probes from teh old probes based on the scripts (v6 and v8 vs. the original probe script)
- LinkerL is 4nt longer (with TATT added to 3' end) for the new probes
- LinkerR is 8nt longer (with CAGGCAGA added to 5' end)
- The UMI sequence is 6nt vs the original 8nt used in the original probe
- For the new probes, the maxH1H2Len parameter is 4nt longer than the original (maxH1H2Len=46). Because the original ppDesigner parameter of H1_plut_H2_Len was original set to 42nt for the original 20140922 probes along with the new filtered probes (which was just a copy and paste of lines from the original probe file), there is a 4nt poly-T filler within every new padlock probe designed
- Based on these differences, it doesn't look like there is anything that would cause drastically worse capture
Verify Raw Probe Sequences[edit]
- We also wanted to verify the raw probe sequences that were sent to Custom Array for order between the two orders (original vs. filtered). Below are the raw files submitted to Custom Array for probe order: (both have multiple probes being ordered, so we need to filter to the correct primer version)
Original 20140922 Probes: File:90k oligos 30Sept2014.txt.gz Filtered Probes: File:CJWEI May2015 12k.txt
- We then want to remove the linker sequences and adapter sequences (AP1/AP2) as listed in the corresponding perl scripts and only extract the two capture arms for either hemo or phusion filtered probe sets. This was done in genemapster at </media/3TB_slot2/cjwei/Fate_Mapping/pp_design/ppDesigner_20140922/regular_probes_noC/filtered_Probes_20150523/verify_probes>
- Based on the results, we see all but one probe for each hemo and phusion filtered probe sets contained within the original probe set. These are contained within the <comm.uniq_new.output> files for each:
Unique to Filtered Probes (Hemo): CCAAATGTATCACGCCCAAAGC AGTGGTGATTCAACTGATTCTGATCGGATACGCATGAGGCT Unique to Filtered Probes (Phusion): TGTCCAGAGGTAGGCTAGGC ACCACTGTTTAGAGACACAGAC
- Also, you'll notice that there are actually 1981 probes designed for each. This is because the "wc -l" linux command does not count the last line, which does not have a return key at the end of it
- Based on this result, it looks like all the probes were ordered correctly except of the one extraneous probe ordered for each set
Comparing UMI Capture[edit]
- Even though we see a dtrastic difference between bias, we also wanted to determine whether at least the trend of the top most captured targets was conserved between the two probe set. Below shows an R image using gpairs in order to do pairwise comparisons of each of the single cell captures. These were all randomly downsampled to the same number of mapped reads per sample with each point representing the number of UMI for a particular target out of 1981 filtered
File:Cw 20150702 Gpairs UMI plot.png
- Based on these plots and correlation, we see that there is very low correlation between any two capture regardless of probes used. However, you'll notice that the plots are skewed towards the original probe axis (even with the non-log scale), suggesting that capture occured better using the original probe set