Redmine

From Clinton:

'Just wanted to touch base and let you know we got our reviewer feedback for the Serratia paper. The reviewers were happy with the paper and gave favourable responses, they have however requested certain amendments before they accept, for publication. Are you able to assist with these before year end?

There are several comments that we need to address and mostly around the genomics. I’ve made a list of things we can do to address the comments with the least amount of work, killing multiple birds with a single change I’m hoping. Not sure if you will have time to do this before you go on leave, but anything you can assist with would be great for Amanda to work on while I’m away from the 9th December.

1. Can we provide genetic context for the contigs which contained resistance genes. The reviewers want more information on what other genes are present on the contigs which contained resistance genes. Perhaps they are on a plasmid or integron with IS elements, or a resistance cluster etc. Is there an easy way to see what other genes are on each contig with a resistance gene identified? We could perhaps narrow this to the beta-lactamase genes/contigs only (TEM_1D, KPC_1, DHA, OXA_1, CTXM_1, SRT_SST, OXA_48, OXA_181, CMY), which is the focus of this paper.'
2. The reviewers would like an indication of the number of SNP differences between isolates. This is a bit odd, as the dendrogram is a representation of that. Could you perhaps provide a table of the count of SNP differences between isolates, and perhaps include a ruler on the dendrogram to indicate SNP difference length. I think we did this for the Pseudomonas paper.
3. For the Full_results figure attached could you please remove the virulence genes (pink) and only have the AMR genes and plasmids. The reviewers would also like to see the gene names more clearly, so if you could adjust spacing, or if its less cramped once the other genes are removed, that would be perfect. The reviewers are also not happy with us calling the control strains “controls” and would prefer we call them Contemporary, if you could please rename those in the legend on this figure. So, isolate 24 would then have to be the same colour as the other Contemporary strains (20, 21, 22).
4. Similarly for the Assembly figure attached here, please rename Control to Contemporary, and make isolate 24 the same colour as 20, 21, 22.
5. One of the tricker comments is why 2 of the resistant isolates did not have carbapenemase genes detected (6-2 has OXA-1, which doesn’t always confer resistance, so we can explain that one), but 25 has no carbapenemase genes detected. This is not uncommon, as they can be resistant due to other mechanisms, but there is one other gene (bla-SME) which is uncommonly identified, but occurs in Serratia, and probably not in the database we used. Is it possible to screen all the isolates for the presence of this gene, I know it’s not advisable to map to short fragments or individual genes, but it’s been done in the attached paper, and we can just cite them? They did the analysis below. Can we map to these 3 accession numbers and see if any isolates have this genomic island (SmarGI1-1), and if so, how many SNP differences there are?

Short-read WGS data were mapped against SME-associated S. marcescens genomic island SmarGI1-1 sequences.3 This revealed that the SmarGI1-1 and flanking regions of isolate 1 were identical to GenBank accession number KF615855, in isolate 4 differ from KF445086 by 8 SNPs and in isolate 5 differ from KF615855 by 53 SNPs.

Individual results will be combined with run1

After these three runs there will be one more run to process. However given time pressure from funders I will process and profile runs 1–4 for now. See Wiki for further details

1. Using the assembled genes compare clinical samples of interest (S21, S22, S23) to controls. We couldn't get stats for S24 (being only 1 sample), just a count difference. S21–23 are non-CRE (carbapenem resistance element) isolates; S24 is an ESBL (extended spectrum beta-lacatamase)

2.Consider further plasmid analysis. It depends what you want from the plasmid analysis – are we just trying to detect known plasmids? In that case srst2 is a good tool to use. If you want to have more info like %coverage and %genes present relative to known plasmid reference sequences then we'd need a different approach (something like PlasmidSeeker)

3.Some of the isolates do not harbour carbapenemases or ESBL genes, but are carbapenem resistant. One of the other mechanisms of resistance are porin changes which prevent entry of the drug into the cell. These are typically mutations to outer membrane proteins like ompC and ompF, which were highlighted in a recent paper. I suspect they were identified using the presence/absence analysis and then investigated further. Would you be able to extract the gene sequences for these genes for us to inspect manually for SNPs which may be causing the observed resistance (do multisequence alignment including reference sequence)? (Note: We did something similar for the Strep pneumoniae project if I remember correctly, so I can modify those scripts to do alignments). The gene list so far is: . AcrB/AcrD/EmrA/Mdt/OmpK36/CpxA/EnvZ/RpoE.

See attached publication

Avril WAlters requested KEGG pathway analysis for this project. Previously I sent them differential gene expression testing results as well as transcription module testing results.

Pathway analysis was conducted using Signaling Pathway Impact Analysis ( [[[https://www.researchgate.net/publication/235418844_SPIA_Signaling_Pathway_Impact_Analysis_SPIA_using_combined_evidence_of_pathway_over-representation_and_unusual_signaling_perturbations#fullTextFileContent]]] )

These results were updated at Stefan's request to include 5 samples previously typed as 'not found' (NF) due to one marker (of 6 total)that differed from the ST303 type. We decided that these are most likely ST303 as it does not match any other known sequence types. They were therefore included with ST303s in the differential fisher's exact testing. This resulted in 51 significant features after multiple testing correction. Results attached.

Metabolomics analysis was conducted by BGI China and the results were transferred by FTP. In addition to the raw and normalized quant data they also performed differential abundance testing, PCA, pathway enrichment etc. Justin Nono has requested that I redo these analyses and integrate with RNAseq profiles. See email correspondence below:
On a follow up note, I have just received the metabolomics results from BGI on the plasma samples (see attached and mail below). Please do not hesitate to retrieve the files , as instructed by the service provider and let us know how, at your convenience, what could be found from the bioinformatic analyses (comparative metabolite expression, untargeted)....I plan to have this done for RNA seq, Metabolomics and the upcoming metagenomics then see if an integration of the 3 platform results is possible to define robust combo of markers with high PPV or strong association with either infection and /or fibrosis.

The comparisons to be done are as follows, where no_merge, merge_D, merge_E refer to the folders made available for transfer from BGI

no_merge:
KK+US- 10 (group 2) vs. KK-US- (group 4)
KK+US+ 10 (group 1) vs. KK-US- (group 4)
KK+US+ 10 (group 1) vs. KK+US- (group 2)

merge_D:
KK+US- KK-US- 20 (group 2 + group 4)
vs.
KK+US+ KK-US+ 20 (group 1 + group 3)

merge_E: metabolites that might regulate infection overall
KK-US- 10 (group 4)
vs.
KK+US+ KK+US- 20 (group 1 + group 2)

I helped Mandy to download her data (SFTP .fastq files from public site at Crick). Some of the files are controls (~500MB) to check if the sgRNA libraries are comprehensive and unbiased while other files contain experimental data (~2GB). Mandy would like to get a quick look at the control files to see if the libraries are good. I will look at running MAGECK on HPC as a nextflow pipeline (docker container available).

After careful assessment of the fairly comprehensive results provided by the collaborators (all done in R), which included differential expression testing between different cell types (microglia, astrocytes, neurons) and treatments (H37Rv, control, BCG) and funtional enrichment analysis, I noticed a batch effect in the data. I will therefore have to redo all analyses after batch correction using Combat.