Tag Archives: troubleshooting

Chloroform Clean Up – Lexie’s QPX RNA from 20110504

After submission of QPX samples to HTGU for Illumina library prep yesterday, I was notified that there was insufficient RNA for the QPX RNA samples. I checked the source RNA on the Roberts Lab NanoDrop1000 and determined that they had high phenol contamination (large peak at 270nm), which results in a large exaggeration in the OD260 absorbance (NanoDrop1000 report[JPEG]; notice terrible OD260/280 ratios; did not save screen shot of absorbance peaks.). As such, the concentrations that Lexie had listed in her notebook for these samples are highly inaccurate and highly inflated. To remove the phenol, I brought all of her QPX RNA samples from 20110504 up to ~200uL with 0.1%DEPC-H2O, added 200uL of chloroform, vortexed for 30s, spun at 12,500g RT for 15mins, and transferred aqueous phase to new tube. Then performed an ethanol precipitation on the aqueous phase. Added 0.1 vols of 3.0M sodium acetate (pH = 5.2), 2.5 vols of 100% EtOH, mixed and incubated at -20C for 1hr. Pelleted RNA by spinning at 16,000g 4C for 15mins.

Results:

As suspected, most of these samples have absolutely no RNA in them. However, the samples that do (the “Control” samples), look great! Pooled 2ug each of the RT Control a & b samples and pooled 2ug each of the 10C Control a & b samples (which are ATCC). Calculations are here. Will take them down to HTGU tomorrow to replace the bad samples that were provided yesterday.

RNA Isolation – C.gigas Larvae from 20110412 & 20110705

RNA was isolated from C.gigas larvae collected from Taylor Shellfish hatchery on the dates noted above. Samples were in RNA Later. RNA Later was removed. Attempted homogenization with a pestle proved futile, as a significant quantity of larvae were sticking to the pestle and were nearly impossible to wash off using TriReagent as a rinsing agent. Due to this, all samples were vortexed for 1min in 1mL of TriReagent. It should be noted that the TriReagent took on a cloudy appearance and even showed some separation into two layers upon letting the samples sit. This was not normal and I was immediately concerned about the high salt content from residual RNA Later. Samples were treated normally with the following changes:

  • Aqueous phase after chloroform treatment was clear, but grey in color. This is not necessarily unusual.

  • Addition of isopropanol triggered immediate precipitation of a dark grey material.

  • “Pelleting” of the RNA after the isopropanol precipitation resulted in a gooey grey material that did NOT pellet, and a clear supernatant. The grey goo was transferred to a clean tube. An additional 500uL of isopropanol was added to the clear supernatant of two samples (#140 & #142), as well as to the grey goo. The addition of isopropanol to the clear supe resulted in an immediate precipitation of white salt-like material. The isopropanol appeared to have no effect on the grey goo. All samples were stored @-20C in their existing conditions until 20120116.

  • Since the two samples that were treated with an additional 500uL of isopropanol produced an excess of salt precipitation, I instead added 1mL of 70% EtOH to all the remaining samples; both the clear supernatants and the grey goo. The idea being that the higher water content in the 70% EtOH would help to keep the salts in solution, while precipitating the RNA. Samples were pelleted. All of the grey goo samples produced a white pellet. The grey goo seemed unchanged. Supernatants (including grey goos) were discarded and the resulting pellets from all samples were washed in this fashion were washed three more times.

  • Pellets were resuspended in 25uL of 0.1% DEPC-H2O and stored @ -80C until 20120123.

  • Samples were spec’d on the Roberts’ Lab NanoDrop 1000.

Results:

Spreadsheet of OD readings is here.

Since samples were split into two (clear supernatant and grey goo), they were kept separate through the remainder of the process. Sample names are appended with “-1″ or “-2″. “-1″ samples are grey goo samples and the “-2″ samples are the clear supernatant samples.

Overall, most of the grey goo samples appear to have produced the highest yields and highest quality of RNA, although this is not true for all of the samples.

qPCR – Emma’s New 3KDSqPCR Primers

Due to previous contamination issues with Emma’s primers, Emma asked me to order new primers, reconstitute them and run a qPCR for her to see if we could eliminate her contamination issues with this primer set. cDNA template was supplied by Emma (from 2/2/11) and was from a C.gigas 3hr Vibrio vulnificus challenge. Samples were run in duplicate, as requested. Master mix calcs are here. Plate layout, cycling params, etc. can be found in the qPCR Report (see Results). Primer set used was:

Cg_3KDSqPCR_F/R (SR IDs: 1186, 1187)

Results:

qPCR Data File (BioRad CFX96)

qPCR Report (PDF)

The negative controls (NTC) are negative, meaning they do not cross the threshold set by the BioRad software. However, there is clearly amplification in the NTCs, but they come up late enough that they do not cross the threshold and, thus, generate a Cq value. Additionally, the melt curve reveals peaks in the NTCs that are at the same melting temperature as the product produced in the cDNA qPCR reactions. This would potentially imply some sort of contamination, as Emma has experienced.

Honestly, I do no think contamination is the problem. I believe that the “contamination” being seen in the NTCs is actually primer dimer. Increasing the annealing temperature (I’m not sure if Emma tried this during her troubleshooting) could potentially alleviate this issue. However, I’m not sure she’s amplifying the target that she wants to. Based on my analysis, I think she needs to re-design primers for her 3KDS target. Read my analysis and why I came to this conclusion below.

It seems unlikely that two independent people (and multiple primer stock replacements!) would have contamination, so I looked in to things a bit further.

I BLASTed the primer sets (NCBI, blastn, est_others db, C.gigas only) and the BLAST results reveal the primers matching with a C.gigas EST sequence that would produce a band of only 63bp. Here’s a screen capture of the BLAST results:

This result does NOT agree with what is entered in our Primer Database. As entered in our sheet, the expected PCR product would be ~102bp. However, taking in to account the BLAST results, it would be difficult to distinguish the difference between primer dimers and PCR product in a melt curve analysis.

Emma has previously run a conventional PCR with these primers and ran a gel (see below). At the time, it was thought to be contamination, but in retrospect (knowing the results of the qPCRs and the BLAST results) it seems likely that what she’s seeing in the negative controls was actually primer dimer, which was the same size of her PCR product (which she thought should be larger). Additionally, the gel was difficult to interpret because no ladder was run. A ladder might have revealed that her PCR product was half the size that she was expecting:

NanoDrop1000 Comparison – Roberts vs. Young Lab

A previous comparison was performed (see 20110209), but it was determined that the standard DNA being used to test the machines was old/degraded. Lisa ordered a new standard DNA dilutions series (Quant-iT dsDNA Kit; Invitrogen) and these DNAs were used. All DNAs were measured 5 times and were mixed by gently flicking between each measurement. A “blank” was measured between each different [DNA] and, if the reading was > + or – 1ng/uL, the machine was reblanked.

Results:

Quick assessment is that Graham’s NanoDrop1000 is more accurate than ours.

Here is a spreadsheet with averages, standard deviations and experimental error (%). Below are the raw measurements from both machines.

Roberts Lab ND100:

Young Lab ND1000:

Data Analysis – Young Lab ABI 7300 Calibration Checks

All runs (3 runs were conducted) were created using a master mix containing C.gigas gDNA (either 50ng or 100ng), 1X Promega qPCR Master Mix, 0.2uM each of forward/reverse primers (18s; Roberts SR ID: 156, 157). The master mix was mixed well and 10uL were distributed in each well of ABI plates. Plates were sealed with ABI optical adhesive covers.

It should also be noted that this analysis was only done with a single primer set and was not tested on any other qPCR machines. This can easily be done if it is desired, however I think one of the issues still being observed with the machine is sample-independent (see Results section below).

Results:

Here’s an extremely quick and dirty analysis of what these qPCR runs have revealed (across the entire plate, 3 plates of data):

Avg. Range of Cts Across Plates – 1.70

Avg. Std. Deviation of Cts Across Plates – 0.352

Based off of the graphs below (particularly the Ct vs Well Position plot), my conclusion is that the machine reads plates inaccurately in Rows A, B, C, F, G, & H. Rows D & E exhibit the most consistent well-to-well readings and, potentially, could be used for qPCR.

The entire work up (which includes a breakdown of each well position relative to each other) is here (Excel Workbook .xlsx). Below are screen captures of one of the three plates (as an example, since all looked the same) that were used for analysis of the amplification plots, melt curves and Ct vs Well Position and a quick description/assessment of what I have observed.

The amplification plot (below) clearly shows the type of spread in Cts across an entire plate that was observed in each run, as well as a large range in fluorescence detected (Rn) in each well.

The melt curve (below) reflects the large range of detected fluorescence seen in the amplification plot. Additionally, some wells exhibit small “bumps” between 75C and 80C. This provides more evidence for a problem with well-to-well consistency.

A graph of Ct vs. Well Position (below) reveals some enlightening information. From looking at this plot, it’s clear that the machine reads from A1 to A12, then B1 to B12 (reads by row, not column) and so on. This plot reveals that most of the variation seen in Ct values occurs in the two rows closest to the edge of the plate, and within those rows, the middle wells’ Cts are more similar to the Cts observed throughout the rest of the plate.

qPCR – Test Young Lab qPCR Calibration

This is a repeat of the two runs from yesterday, just to see if there is a correlation between the failed plates being the first of the day or not. Master mix calcs and cycling params are here (these calcs are from yesterday, but were used again for today).

Results:

Amplification in all wells. Still seeing ~3 cycle spread across the entire plate. Will work up all three successful sets of run data.

NanoDrop1000 Comparison – Roberts vs. Young Lab

Due to an apparent reduction in assay sensitivity for the Hematodidium qPCR assay, we have decided to determine if the spec readings of the plasmid DNA being used for the standard curves are accurate. Used C.gigas gDNA and the lambda DNA Standard (100ng/uL) included in the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen) that was marked as received 9/1/10. Tested both the Roberts Lab and Young Lab using these DNAs. At least 6 separate measurements were taken of each DNA on each machine. Samples were briefly mixed by flicking the tube 4-5 times prior to each measurement.

Results:

Spreadsheet containing absorbance data and calculations of average concentration and standard deviation for both DNA samples on both machines is here.

A quick table of the results:

Roberts Lab Young Lab
[Avg. gDNA] (ng/uL) 45.656 51.778
Std Dev gDNA 0.2377 0.5825
[Avg. lambda DNA] (ng/uL) 76.01 90.255
Std Dev lambda DNA 3.826 0.9342

The first thing to notice is that the lambda DNA that has been used for standard curves does not have the expected concentration (100ng/uL) on either of the machines. It seems unlikely that BOTH NanoDrop1000s are incorrectly calibrated (which could be a possible explanation for why the lambda DNA is not matching the expected 100ng/uL). This is also supported by recent curves done on the Friedman Lab plate reader using this DNA by Lisa, Vanessa and Elene (data not shown). It’s also interesting to note that the lambda DNA also shows a greater standard deviation (on both machines) than the other DNA (gDNA) used in this test. This is surprising as one would expect a store-bought reagent to be of the highest quality, particularly since it is supposed to be used for DNA quantification. However, it should also be remembered that this DNA is over a year old and has never been aliquoted. As such, it has gone through an extremely high number of freeze/thaw cycles which could have an impact on the long-term quality of the DNA.

The second thing to notice is that the Roberts Lab and Young Lab machines provide different concentrations of each of the two DNAs. Unfortunately, due to the fact that the lambda DNA concentration is not as expected (100ng/uL) on either machine it is impossible to determine which machine is more accurate. However, it appears that the Young Lab NanoDrop1000, overall, is more consistently precise in its readings than the Roberts Lab NanoDrop1000. Of course, both machines do seem to be sufficiently precise that precision shouldn’t be a concern.

I’ve notified Lisa of the potentially inaccurate readings of the lambda DNA. She has ordered a fresh set of DNA standards that will be used to test both machines to help assess their accuracy.

qPCR – Test Young Lab qPCR Calibration (Repeat)

This was repeated from earlier today due to the failure of the previous run, but had to use new gDNA since I ran out of the stock I had previously used. Master mix calcs and cycling params are here.

Results:

Amplification in all wells, however well E4 appears to have had some evaporation (and the effects can clearly be seen in the amplification plot below). Still getting ~3 cycle spread across the entire plate, which is disconcerting. Oddly, this is the second day where the 1st run completely failed, but the 2nd run was successful…

qPCR – Test Young Lab qPCR Calibration (Repeat)

This is a repeat of a run from 20110204. Here’re master mix calcs. This was being repeated to evaluate whether or not the relative differences in Ct values observed on 20110204 are consistent or not across the plate. Cycling params were as follows:

  • 95C – 10min

40 Cycles of:

  • 95C – 15s
  • 55C – 30s
  • 72C – 1m

Melt curve.

Results:

Absolutely no amplification of any kind! Bizarre. Will repeat.

qPCR – Test Young Lab qPCR Calibration (Repeat)

This is a repeat of an earlier run from today, but with a different qPCR plate. Here’re master mix calcs (bottom half of page). Cycling params were as follows:

  • 95C – 10min

40 Cycles of:

  • 95C – 15s
  • 55C – 30s
  • 72C – 1m

Melt curve.

Results:

All samples amplified and showed a proper dissociation curve. However, it does look like there’s a spread of ~3 Cts across the plate. This is not good, as this is the equivalent of ~10-fold difference in gene expression. Will repeat again and see if specific wells show the same relative differences in Cts.