Period From 7/2020 Through 12/2020
The group held a conference call on 7/24 and 9/4, and 10/16 and one is scheduled for 11/20
July 24: Dr. Testerman reported that she has continued with her work on growing cultures. Her technician is participating in this effort. Dr. Testerman plans to submit a new NORD proposal and she has submitted a proposal focused on personalized medicine. She also discussed some of the compounds that she has been able to isolate from her cultures. Dr. Testerman has continued purifying these compounds and studying their biological activity.
Prior to the conference call, Dr. Merrell emailed a link to a program that his former student Ryan developed to help sort out which ASV’s should be kept for our PMP microbiome research. The number of reads achieved in our microbiome experiment was low and contamination is a real issue. Ryan’s program aims at helping us draw the line on what results need to be retained and what results are caused by contamination. The slide bars on the left of the screen in Ryan’s program can be adjusted and based on the adjustments the plots on the right are altered. The upper plot compares the prevalence of a species in real samples and in negative controls which are the contaminant controls that did not have any tissue samples associated with them. The lower plot compares the average ASV/sample for real samples and negative controls. The blue dots in the plot correspond to specimens that would be kept and the red dots would be considered contamination and removed based on the settings established with the slide bars. As the sliders change, the blue dots can change to red and vice versa. Holding the cursor over a dot produces a label for its bacterial genus. The last plot generated by the program is a genus plot. This plot shows results for non-PMP peritoneal control tissue, PMP tissue from females (F) and Males (M), the mock community, negative controls and one ovarian cancer specimen. The mock community is a defined mixture of microbial cells and/or viruses or nucleic acid molecules created in vitro to simulate the composition of a microbiome sample or the nucleic acid isolated therefrom. A long discussion about the results shown in Ryan’s program took place. In order to proceed to the next step it was suggested the members of the group individually run Ryan’s simulation and develop guidelines on where to draw the line for determining what to remove as a contaminant. Member’s suggestions can be communicated via email to the group before our next conference call.
Ms. King reported that Mercy will be adding a new surgical oncologist in the coming months. The person has expressed an interest in research. Mercy personnel have continued to work remotely but patients are still being enrolled. Ms. King will be adding Dr. Metcalf to Mercy’s PMP protocol as a new researcher. She will also be checking what will be necessary for PMP specimens to be sent to Dr. Metcalf and whether a separate MTA agreement is required.
Sept. 4: Dr. Testerman reported that she has submitted a new proposal to NORD as well as a poster for a NORD event. She has continued her culturing of tumor cells. Dr. Testerman noted that some cultures take a very long time to grow. In this regard she mentioned a DPAM culture that produces a lot of mucin. Dr. Testerman’s program of purifying and testing various compounds isolated from her bacterial cultures is continuing. She noted that some of the compounds affect oxygen consumption, but she was unsure if cell growth rate and oxygen consumption were correlated. In addition to testing the compounds on cancer cells she is also testing them on normal cells. Some of the compounds she has found are known and one compound is totally new. Interestingly, compounds resulting from PMP191F are similar to those found in plants. She is interested in finding compounds that can stimulate growth. Dr. Testerman has purchased a new fluorescence microscope and hopes that using it will increase assay throughput. Next, Dr. Testerman discussed her collaboration with GE Sciences. Prior to the conference call Dr. Testerman emailed a Multiplex data summary of results achieved on PMP specimens. GE Sciences has an H. pylori antibody that they use but they are unwilling to identify the peptide that the antibody recognizes. In order to make results for H. pylori more accurate it is proposed that CagA can be added to the analysis. The images that were emailed were discussed. The slides show results for 8 different markers each with a different color. Dr. Testerman has been in touch with Dr. Studeman from Mercy to discuss the GE Sciences results.
Dr. Merrell reported that he had received some input on Ryan’s program from the group but that not much progress has been made in deciding which microbiome specimens should be eliminated. Ryan’s job has kept him very busy of late and he has not been able to devote time to tweaking the program he developed for us. Dr. Merrell also discussed the issue of sending specimens to Dr. Metcalf for an independent microbiome analysis. Ms. King was able to get Dr. Metcalf added to Mercy’s protocol. Since USUHS has the PMP specimens a MTA agreement between USUHS and Colorado State needs to be executed. Dr. Merrell has initiated the process of securing the MTA. Dr. Metcalf mentioned that she has 2 students who could devote time to the analysis. Dr. Merrell mentioned that he has ~30 PMP specimens that were already used for microbiome sequence analysis that can be sent to Dr. Metcalf since some of the PMP specimens he had have been exhausted.
Ms. King reported that Mercy will be back in the office part time next week. She would like to send specimens from new patients to Dr. Testerman for inclusion in the GE Sciences study. Next a discussion took place about the PMP genome study taking place at Mercy. Mercy is working with a private contractor who is charging $900/specimen. Due to the cost, only 6 cases have been analyzed so far. Additional cases including signet ring cell tumors will be studied in the future. Dr. Testerman mentioned that doing a genome study on cultured cells would be easier than on tumors themselves since the tumors require micro-dissecting. Ms. King said that she will email genomic results obtained so far to the group. Dr. Merrell mentioned that a new DOD request for proposals on rare cancer has been published. This DOD initiative could allow for joint genome research between our group and Mercy. Dr. Merrell will look into the details about this RFP and what is required for a pre-proposal.
Oct. 16: Dr. Testerman reported that she has submitted a new proposal with GE Sciences with input from Ms. King and Dr. Sardi whom she thanked. No new animal experiments or culturing experiments have been carried out since the last call. Dr. Testerman continues testing compounds resulting from her cultures and several have been toxic to tumor cells. Dr. Testerman has also begun comparing cell lines for their susceptibility to mitomycin c treatment.
Dr. Merrell asked Dr. McAvoy to report on his calculations using the decontam program on our microbiome data. This program aims at detecting contaminants in microbiome data using a statistical approach. Prior to the call Dr. McAvoy emailed a discussion about applying decontam to our microbiome data as well as 2 plots, and 2 spreadsheets with results and a copy of the paper describing the decontam to the group. The text of this email is below.
This email is a modified version of my earlier email on the decontam results. I ran the prevalence decontam program (IsContaminant) on our PMP microbiome data and have attached my results as 2 plots and 2 Excel tables. I ran 2 cases: PMP versus Neg-Ctrls, and Controls versus Neg-Ctrls. A description of this program is given in the paper1 discussing decontam as follows:
Statistics Used: For each sequence feature, a chi-square statistic on the 2×2 presence-absence table in true samples and negative controls is computed, and a score statistic P is defined as the tail probability of the chi-square distribution at that value. The p value from Fisher’s exact test is used as the score statistic instead if there are too few samples for the chi-square approximation. The score statistic ranges from 0 to 1. Small scores indicate the contaminant model of higher prevalence in negative control samples is a better fit. Although the prevalence-based score statistic is set equal to a p-value from the chi-square or Fisher’s exact tests, it is used by decontam only as a score that effectively distinguishes between the contaminant and non-contaminant mixture components. This treatment is also recommended by the potential for cross-contamination to violate distributional assumptions related to independence between samples.
Method: Prevalence-based contaminant identification remains valid in very low biomass environments where a majority of MGS sequences might derive from contaminants rather than true inhabitants of the sampled environment (i.e. Contaminant (C) ~ True Sample (S) or C>S). Even in the low-biomass regime, it is still expected that non-contaminants will appear in a larger fraction of true samples than in negative control samples.
The program uses a default probability threshold of p=0.1 to classify contaminants.
PMP versus Neg-Ctrls Results: Of the 721 measured ASV’s the program detected 24 as contaminants. The attached plot shows the prevalence of true samples (red) versus the prevalence of contaminants (blue). Each circle gives the number of times an ASV was detected in a negative control and a true sample. I suspect that some blue circles have been covered by red circles since there are less than 24 blue circles. In the attached Excel file the ASV’s identified as contaminants are shown in red (opposite color from plot) together with their calculated probability. The IsContaminant program also allows calculating which ASV’s are more prevalent in negative controls than in true samples. In the Excel file these ASV’s are shown in the green rows which should be added to the red rows. All the red values were more prevalent in the Neg-Ctrl’s than in the PMP samples. The sum of the red and green rows is 100, 24 contaminants (red) and 76 true samples (green). In looking at the attached plot for this case I noticed that there were points at 0,0, which indicates that an ASV was not measured in either the PMP samples or the Neg-Ctrls. I checked the raw data and found that there were 61 such ASV’s and they are listed in the blue rows in the spreadsheet. Thus it looks like the 721 ASV’s reduce to 721-24-61=636 ASV’s associated with the PMP samples.
In running the decontam program with the PMP data I noticed something about the PMP data. Of the 721 ASV’s, 511 had only a single value which was measured in only 1 of the 75 samples, none of which was a negative control. The decontam program in this case cannot calculate a probability. The cases with a single ASV read were determined to be true samples by the program. All of the negative controls had more than 1 ASV read in their data.
Control versus Neg-Ctrls Results: Of the 721 measured ASV’s the program detected 3 as contaminants. The attached plot shows the prevalence of true samples (red) versus the prevalence of contaminants (blue). Each circle gives the number of times an ASV was detected in either a negative control or a true sample. In the attached Excel file the ASV’s identified as contaminants are shown in red (opposite color from plot) together with the calculated probability. Two of the three contaminants are the same as for the PMP Neg-Ctrls case and the third is different. The calculations showed that in addition to the 3 contaminants, 37 non-contaminant ASV’s were also more prevalent in the Neg-Ctrls than in the Controls. These 37 ASV’s are listed in the green rows in the Excel table. In looking at the attached plot I again noticed that there were points at 0,0, which indicates that an ASV was not measured in either the Controls or the Neg-Ctrls. I checked the raw data and found that there were 519 such ASV’s and they are listed in the blue rows of the spreadsheet. Thus it looks like the 721 ASV’s reduce to 721-3-519=199 ASV’s associated with the Control samples.
Although not mentioned in this email the black entries in the spreadsheets give real data for either PMP ASV’s or Control ASV’s where the negative controls are less prevalent than the true samples. A discussion about the decontam results took place. The issue concerning ASV’s with a single read, i.e. a singleton, was discussed. Dr. Johnson suggested that in addition to prevalence, abundance be considered. Dr. Merrell suggested that the spreadsheets that were emailed could be merged with another spreadsheet that Dr. Johnson had developed so that both prevalence and abundance could be seen in the same spreadsheet. Dr. Johnson agreed to do the merge. A discussion about subtracting contaminants took place. Dr. Metcalf suggested that it would be interesting for our analysis to add in microbiome results that she will generate on our PMP specimens. The paper work for shipping the specimens from USUHS to Colorado State is in place and Dr. Merrell’s lab manager will follow up and coordinate the shipping. Dr. Metcalf said that there is support for translational research that she may be able to tap into to pay for the microbiome analysis.
Ms. King reported that there are no major updates from Mercy. A discussion took place about submitting a pre-proposal to the CDMRP initiative on rare cancers. Mercy in interested in DNA and RNA sequencing of PMP tumor specimens. The pre-proposal which is due Nov. 12 will be used by the CDMRP to invite submission of full proposals. There were two possibilities for pre-proposal submissions that were discussed. One would involve individual submissions from Mercy and Dr. Testerman. The other would be a joint submission addressing DNA sequencing, RNA sequencing, bacteria culturing and staining for studying PMP. Dr. Merrell thought that a joint submission would be stronger. The fact that collaboration on PMP has been going on for a number of years would be a strength for a joint submission. Ms. King and Dr. Testerman agreed to follow up on how to proceed with a pre-proposal.
Nov. 20: Dr. Testerman reported that in her culturing work, the cells with mesenchymal phenotype grow slowly. Those that look more epithelial grow more quickly, but still not as fast as most cell lines. Dr. Testerman will submit a grant proposal to look at compounds that have resulted from her culturing work. This proposal will be a rewrite of an earlier proposal to NIH, and it will be submitted to the DOD rare cancer proposal RFP. Dr. Testerman also reported that her proposal involving personalized medicine was not funded. She plans to team up with a cancer biologist and resubmit the proposal. Dr. Testerman has gotten a whole genome sequence for two of the bacteria she cultured from PMP tissue. One of these bacteria is probably a new species. This bacterium does not like to use sugars for growth, but prefers amino acids. One of Dr. Testerman’s cultures from PMP 498 yielded an interesting compound that preliminary data shows is cytotoxic for PMP cells. Dr. Testerman has submitted a letter of interest to NORD for their 2020 PMP funding program and is waiting to hear if she will be invited for a full application.
Dr. Merrell asked Dr. Johnson to discuss his revised microbiome program. The program has 3 plots that can be used to help eliminate contaminants in our microbiome data: 1. Percent Prevalence, 2. Average Read Abundance, and 3. Genus (top 10) Relative Abundance. Sliders at the top of the program can be adjusted and the effect on the three plots can be seen. In the first 2 figures the red circles show species that should be removed and considered contaminants, the blue circles are considered real data. The transparent circles give the contaminants that were found by the decontam program. In both the first 2 plots, data for the mock ASV’s has been eliminated. Real samples include PMP and control ASV’s. The x axis in both plots is negative controls. If the computer mouse is held over a circle its genus is shown next to the circle. It was suggested that the group work with Dr. Johnson’s program so that we can better define what should be eliminated as contaminants, assuming that we eventually publish our results. A discussion about the microbiome results shown in the plots took place.
Ms. King reported that due to an increase in covid cases at Mercy their operations have been significantly scaled back. As a result no new samples have been shipped to Dr. Testerman. Once the covid spike ameliorates, sample shipment will resume.
The MTA agreement between USUHS and Colorado State has been finalized and the lawyers at both institutions have signed off. Dr. Metcalf will apply for pilot grant funding that is translational in scope. She will look into measuring the microbiome in PMP specimens plus specimens from three other types of cancer. Her results can be added to the database that we already have, and help in determining what is a contaminant and what is a real species in the microbiome.