Transform and library prep fk8 DMS library

[kychen37]

GanttStart: 2023-04-20

Background

In #3 and https://github.com/rasilab/rqc_aggregation_aging/issues/101, we identified and validated FK as a stalling dipeptide.
In https://github.com/rasilab/rqc_aggregation_aging/issues/117 we identified other possible targets of RQC (FK and others) using the 8x dicodon (pHPSC1142) library in hel-del cells.
In https://github.com/rasilab/rqc_aggregation_aging/issues/119, we identified endogenous genes with FK/etc stalling type motifs.
In https://github.com/rasilab/rqc_aggregation_aging/issues/120 we designed libraries to test these endogenous motifs and DMS libraries
In https://github.com/rasilab/rqc_aggregation_aging/issues/121 we cloned these libraries as a pool and in https://github.com/rasilab/rqc_aggregation_aging/issues/122 I transformed/library prepped them.
In https://github.com/rasilab/rqc_aggregation_aging/issues/124 we analyzed the deepseq data and found that the FK8 DMS library had very little representation in the library.
In https://github.com/rasilab/rqc_aggregation_aging/issues/129 we re-cloned the fk8 library alone (oKC224) into the library reporter without barcodes (pHPSC1163). Colony PCR + sangseq indicates that only ~12.5% of this library can be expected to be correct, so 6-8K (over 1184 variants) while 40-60K are either parent or have an incorrect insert.

Here I will tranform into WT and Hel2-del yeast and prepare libraries to redo this sequencing experiment.

Strategy

• High-efficiency transform BamHI+Not1-cut pHPSC1163 into WT and Hel2-del yeast with liquid recovery.
  • 2x50ug plasmid into 2x1e9 cells -> ~20,000 integrants each for WT and Hel2-del
• Do library prep as usual as per below:
• Select in 100mL SCD-URA for 72hrs total, passaging 100 OD*mL cells (1 billion cells) in 100mL SCD-URA every 24hrs. Make glycerol stocks.
• Inoculate from frozen glycerol stock to 50mL YPD for starting OD of 0.5 (higher than usual to compensate for cell growth coming out of frozen), grow overnight
• Dilute to OD 0.1 for mid-log phase harvesting
  • Must include scKC190 and scKC191 spike-ins at this point, because comparing across WT and Hel2-del strains did not work well using either RPM or median-normalization in https://github.com/rasilab/rqc_aggregation_aging/issues/124. Spike in at 10/1024
• Harvest many 50mL OD 0.4-0.5 pellets (200million cells/pellet)
• Extract RNA from 200 million cells (expect >=10ug) and gDNA from ~400 million cells (expect >=2ug)
  • got less than expected, 0.9-2ug yield
• IVT from 750ng gDNA (75million cells)
• RT using 1/2 of the yield for mRNA and gRNA with oKC235 UMI primer
  • 5ug input for RNA samples, 7ug input for gRNA samples
• One round of PCR using oPN776 and oKC230-240 index primers
• Sequencing with oKC236 custom read2 and either oKC237 custom or standard read1.
  • Note that spikeins will work with the custom read2 sequencing primer, but will not work with custom read1 sequencing primer (which is fine, I just need it in one direction anyway).

Experiment Links

first replicate: https://github.com/rasilab/rqc_aggregation_aging/blob/master/experiments/kchen_exp74_redo_fk8_dms_deepseq.md

second replicate: https://github.com/rasilab/rqc_aggregation_aging/blob/master/experiments/kchen_exp75_redo_fk8_dms_deepseq_rep2.md

library prep: https://github.com/rasilab/rqc_aggregation_aging/blob/master/experiments/kchen_exp76_library_pre_fk8_dms_replicates.md

snapgene map: https://github.com/rasilab/snapgene_maps/blob/master/DNA%20Files/lab_database/kchen/illumina_amplicons/IKCSC12_R1_fk8_umi_R2.dna

Brief conclusion

In the analysis we felt there was evidence of RT/PCR bias because T/C containing codons were preferentially low in some of the conditions. Other conditions did not have this bias and looked good. I tried redoing the library prep using the correct primer (https://github.com/rasilab/rqc_aggregation_aging/issues/135#issuecomment-1597885161, https://github.com/rasilab/rqc_aggregation_aging/issues/135#issuecomment-1603766677), but the RT and noRT samples no longer separated for most samples. Since the mRNA and gRNA samples are prepped the exact same way with regards to RT and PCR, any bias should normalize out, so we decided not to redo and kept the data from this run.

Checklist before closing issue

• Are all plasmids, oligos, and cell lines in their correct location?
• Is the plasmid / cell line entered into our lab database?
• Is the plasmid map given appropriate name and moved to the lab_database folder on Snapgene?
• Is the plasmid map pushed to https://github.com/rasilab/snapgene_maps/?
• Is the lab notebook link up-to-date without any broken links to images?
• Are there any intermediate plates or reagents that need to be disposed off?

[kychen37]

I was thinking about our conversation re replicates and I think it's worth doing 2 more biological replicates (aka two more transformations) so there are 3 total replicates. That will be 12 libraries total and ~60million reads, let me know if you don't think it's worth it @rasi

[rasi]

@kychen37 Yes, that is a good idea if it is not too much work.

[kychen37]

Update

@rasi
I plan to library prep both replicates next week following strategy outline above, 4 libraries per replicate x2 = 8 libraries total. These should be ready by end of next week.

[kychen37]

Update

@rasi
The cDNA samples that had RNA as input took more diluting to get reasonable CTs than I was expecting (usually a 1:8-1:16 dilution is good enough for getting CTs <20, but this time requires at least a 1:32 dilution). With this dilution the PCR reaction will be very large, so I have to go through the logistics of the library prep in more detail, so I think these will actually be done closer to end of next week

[rasi]

@kychen37 I am not sure I understand the comment above. Does this mean you have more RNA going as input into RT, in which case you can get away with smaller PCR reactions?

[kychen37]

Update

@rasi

• PCR using 10% of RT reaction for RNA samples and 20% of RT reaction for gRNA samples -> RNA samples reached CT in 17 cycles, gRNA samples reached CT at 10 cycles
• cleaned up between 50-80% of each PCR sample (lost some volume bc I accidentally sloshed liquid around in the column and decided to toss those columns entirely to avoid cross-contamination)
• gel-extracted -> submitted for tapestation
• when tapestation results return I wiill know if these libraries are ready for submission
• 8 total libraries

[kychen37]

Update

@rasi final libraries:

label	cDNA sample	RNA	cycles	variants	target/variant	target	ng/uL
76lib1	76cDNA1	WT RNA	17	1024	10,000/variant	10 million	5.76
76lib2	76cDNA3	hel2 RNA	17	1024	10,000/variant	10 million	4.42
76lib3	76cDNA5	WT RNA rep2	17	1024	10,000/variant	10 million	4.2
76lib4	76cDNA7	hel2 RNA rep2	17	1024	10,000/variant	10 million	2.46
76lib5	76cDNA9	WT gRNA	10	1024	10,000/variant	10 million	0.634
76lib6	76cDNA11	hel2 gRNA	10	1024	10,000/variant	10 million	0.848
76lib7	76cDNA13	WT gRNA rep2	10	1024	10,000/variant	10 million	0.878
76lib8	76cDNA15	hel2 gRNA rep2	10	1024	10,000/variant	10 million	0.95

I was initially concerned that the double band in the tapestation might be incomplete libraries, but even when I tried doing a final extension of 15m at 72C the double band persisted. The double bands are close enough in size that the actual tapestation quantification doesn't say there are two bands (which it did when there were incomplete libraries in my previous runs). So I don't think the double bands are incomplete libraries, they are likely the spike-ins I used, since the 48nt inserts in these spikeins were truncated they are slightly smaller than the real library. Because there are only 1024 variants in this library, I spiked in at higher concentration than I previously have (1/1024 vs 1/200,000), which is why I think they are showing up on tapestation.

Overall I think these libraries are ready to sequence. https://github.com/rasilab/protocols_tutorials/issues/80#issuecomment-1569408268