Jul 26, 2023·edited Jul 26, 2023Liked by Mass ____
Daoyu has been saying that Csabai's Antarctic reads contain C29095T, but actually there's just 3 reads with C29095T, and the bases with the mutation all have a very low quality: quality 6 in base 137 of read 5599415 in SRR13441704_2 (where there's 14 remaining bases after it in the read out of which half are different from the reference), quality 5 in base 94 of read 21585669 in SRR13441708_2 (where 2 bases before it there's another base that is different from the reference that has a quality of 10), and quality 5 in base 2 of read 23910570 in SRR13441708_2 (where the first base of the read is also different from the reference). And a Phred33 quality of 5 corresponds to an expected error rate of 1/10^.5, or about 32%. So I think the 3 instances of C29095T in the Antarctic reads are just sequencing errors.
`samtools mpileup` doesn't include bases with quality below 13 by default, so in order to see the 3 mutations, you have to remove the quality cutoff with `samtools mpileup -Q0`:
for x in SRR134417*.fastq.gz;do bowtie2 -p3 -x sars2.fa -U $x --no-unal|samtools sort -@2 ->${x%%.*}.sars.bam;done
for x in {00..11}_{1,2};do samtools mpileup -f sars2.fa -Q0 SRR134417$x.sars.bam|ruby -alne'next unless[29095].include?($F[1].to_i);next if$F[3]==0||$F[4]=="*";t=$F[4].upcase.gsub(/[.,]/,$F[2]).gsub(/[+-]\d+[A-Z]+/,"").chars.tally;puts [$F[1],$F[2],t["A"]||0,t["C"]||0,t["G"]||0,t["T"]||0]*" "'|sed s/^/$x\ /;done|sort -nk2|(echo run pos ref A C G T;cat)|column -t
The sample Guangdong/FS-S30-P0052/2020 is also interesting, because it has only 5 mutations from Wuhan-Hu-1 but all of them are shared with RaTG13. The sample has all 3 proCoV2 mutations and the 2 additional mutations C21711T and C24382T. However the Antarctic reads don't seem to include either of the additional mutations. In a set of 719,679 GISAID sequences with collection date in 2020 or earlier, if sequences with over 100 mutations are excluded which are mostly other SARS2-like viruses, then C21711T is found in 128 sequences and C24382T is found in 1,156 sequences: `curl -s https://cdn.discordapp.com/attachments/1093243194231246934/1129149184939929713/early.comb.xz |gzip -dc|sort -k4>early.comb;awk -F\\t '$11<100' early.comb|cut -f12|grep -o '[A-Z]21711[A-Z]'|sort|uniq -c|sort`.
For some reason many of these interesting mutations seem to pop up at Guangdong before Hubei, even though GISAID has much more early samples from Hubei than Guangdong. On GISAID the 7 C29095T samples with the earliest collection date are also found at Guangdong: `grep C29095T early.comb|grep Human|cut -f1,3-7,11,12-|head|tr \\t \|`.
In Jesse Bloom's paper about the recovered deleted sequences, in the tree where he selected WA1/proCoV2 as the root, there's only 3 first-level branches under the root, out of which one branch consists of the mutations C865T and C13694T. Both mutations are common in the Antarctic reads, but there's only 4 samples on GISAID with a collection date in 2020 that contain both mutations: 2 samples from Singapore from late January and 2 samples from Guangdong from early February. All 4 samples have only 5 mutations from Wuhan-Hu-1, where the other 3 mutations are the mutations of proCoV2/WA1. I came up with the term "Guangdong-Singapore line" to refer to the line with the C865T and C13694T and mutations. Neither of the mutations seems to be ancestral though. But it's still interesting that the mutations are found in early Chinese samples without C18060T, so there seems to be an overlooked sublineage of proCoV2 which is also found in the Antarctic reads.
When I did variant calling for the pooled reads from all Antarctic runs, I got a high MAPQ value for the mutation A16156G, which is included in only one sample at GISAID with a collection date in March 2020 or earlier, which is a sample from Guangdong with a total of 4 mutations relative to Wuhan-Hu-1. Two of the additional mutations are the lineage A mutations, and the fourth one is T9013A, which is a rare mutation which is not found in any other samples at GISAID until a sample from Brazil with a collection date in May 2020.
C23525T is another rare mutation in the Antarctic reads that got high MAPQ for my pooled reads. It's found in 6 samples at GISAID with a collection date in February 2020 or earlier, but one of the samples is from Guangdong, and its only other mutation relative to Wuhan-Hu-1 is C13694T which is one of the two "Guangdong-Singapore" mutations.
Out of the Antarctic mutations with a fairly high MAPQ, the only mutations which appear to be ancestral are the 3 proCoV mutations, but other mutations appear to not be ancestral: T9440A, C13694T, A16156G, A17039G, A18082G, A18082G, A21975C, C23525T, G23525T, C25498T, C26895T, and G29449T. So I think it's possible that the Antarctic reads don't actually contain strains that are more ancestral than WA1.
When I ran `samtools mpileup` for the Antarctic reads with the default quality cutoff of 13, I got 20 reads which included position 3037 but zero of them had C3037T.
It does not have the slow curve up-front: a toxic batch has a massive up-front effect, which then fizzles off. The Toxic Oil was identified years later.
Yes EVALI was cover-up for early COVID.
We note who insists that it's not, with a level of scrutiny/exploration/data much different from what is applied to WIV/Wuhan-CDC or even Fauxi/Sadsack/Baryc.
What's at play here? Directed Limited Hang-out?
Who died unexpectedly, when?
When did it spread?
- 2009 "Swine" H1N1 Flu is clearly high & out of sync/very early.
Was that a rehearsal, or also a leak? a first attempt *also with a pandemic declaration*, but that fizzled too fast, and also a deadly vakscene withdrawn after maiming a couple hundreds?
The next big ILI curve is 2017-2018 cyan, better synchronized - while it was NOT in Europe
Daoyu has been saying that Csabai's Antarctic reads contain C29095T, but actually there's just 3 reads with C29095T, and the bases with the mutation all have a very low quality: quality 6 in base 137 of read 5599415 in SRR13441704_2 (where there's 14 remaining bases after it in the read out of which half are different from the reference), quality 5 in base 94 of read 21585669 in SRR13441708_2 (where 2 bases before it there's another base that is different from the reference that has a quality of 10), and quality 5 in base 2 of read 23910570 in SRR13441708_2 (where the first base of the read is also different from the reference). And a Phred33 quality of 5 corresponds to an expected error rate of 1/10^.5, or about 32%. So I think the 3 instances of C29095T in the Antarctic reads are just sequencing errors.
`samtools mpileup` doesn't include bases with quality below 13 by default, so in order to see the 3 mutations, you have to remove the quality cutoff with `samtools mpileup -Q0`:
brew install parallel sratoolkit bowtie2 samtools
printf %s\\n SRR134417{00..11}|parallel -j12 fastq-dump --split-3 --gzip
curl 'https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=nuccore&rettype=fasta&id=MN908947.3' >sars2.fa;bowtie2-build sars2.fa{,}
for x in SRR134417*.fastq.gz;do bowtie2 -p3 -x sars2.fa -U $x --no-unal|samtools sort -@2 ->${x%%.*}.sars.bam;done
for x in {00..11}_{1,2};do samtools mpileup -f sars2.fa -Q0 SRR134417$x.sars.bam|ruby -alne'next unless[29095].include?($F[1].to_i);next if$F[3]==0||$F[4]=="*";t=$F[4].upcase.gsub(/[.,]/,$F[2]).gsub(/[+-]\d+[A-Z]+/,"").chars.tally;puts [$F[1],$F[2],t["A"]||0,t["C"]||0,t["G"]||0,t["T"]||0]*" "'|sed s/^/$x\ /;done|sort -nk2|(echo run pos ref A C G T;cat)|column -t
The sample Guangdong/FS-S30-P0052/2020 is also interesting, because it has only 5 mutations from Wuhan-Hu-1 but all of them are shared with RaTG13. The sample has all 3 proCoV2 mutations and the 2 additional mutations C21711T and C24382T. However the Antarctic reads don't seem to include either of the additional mutations. In a set of 719,679 GISAID sequences with collection date in 2020 or earlier, if sequences with over 100 mutations are excluded which are mostly other SARS2-like viruses, then C21711T is found in 128 sequences and C24382T is found in 1,156 sequences: `curl -s https://cdn.discordapp.com/attachments/1093243194231246934/1129149184939929713/early.comb.xz |gzip -dc|sort -k4>early.comb;awk -F\\t '$11<100' early.comb|cut -f12|grep -o '[A-Z]21711[A-Z]'|sort|uniq -c|sort`.
For some reason many of these interesting mutations seem to pop up at Guangdong before Hubei, even though GISAID has much more early samples from Hubei than Guangdong. On GISAID the 7 C29095T samples with the earliest collection date are also found at Guangdong: `grep C29095T early.comb|grep Human|cut -f1,3-7,11,12-|head|tr \\t \|`.
In Jesse Bloom's paper about the recovered deleted sequences, in the tree where he selected WA1/proCoV2 as the root, there's only 3 first-level branches under the root, out of which one branch consists of the mutations C865T and C13694T. Both mutations are common in the Antarctic reads, but there's only 4 samples on GISAID with a collection date in 2020 that contain both mutations: 2 samples from Singapore from late January and 2 samples from Guangdong from early February. All 4 samples have only 5 mutations from Wuhan-Hu-1, where the other 3 mutations are the mutations of proCoV2/WA1. I came up with the term "Guangdong-Singapore line" to refer to the line with the C865T and C13694T and mutations. Neither of the mutations seems to be ancestral though. But it's still interesting that the mutations are found in early Chinese samples without C18060T, so there seems to be an overlooked sublineage of proCoV2 which is also found in the Antarctic reads.
When I did variant calling for the pooled reads from all Antarctic runs, I got a high MAPQ value for the mutation A16156G, which is included in only one sample at GISAID with a collection date in March 2020 or earlier, which is a sample from Guangdong with a total of 4 mutations relative to Wuhan-Hu-1. Two of the additional mutations are the lineage A mutations, and the fourth one is T9013A, which is a rare mutation which is not found in any other samples at GISAID until a sample from Brazil with a collection date in May 2020.
C23525T is another rare mutation in the Antarctic reads that got high MAPQ for my pooled reads. It's found in 6 samples at GISAID with a collection date in February 2020 or earlier, but one of the samples is from Guangdong, and its only other mutation relative to Wuhan-Hu-1 is C13694T which is one of the two "Guangdong-Singapore" mutations.
Out of the Antarctic mutations with a fairly high MAPQ, the only mutations which appear to be ancestral are the 3 proCoV mutations, but other mutations appear to not be ancestral: T9440A, C13694T, A16156G, A17039G, A18082G, A18082G, A21975C, C23525T, G23525T, C25498T, C26895T, and G29449T. So I think it's possible that the Antarctic reads don't actually contain strains that are more ancestral than WA1.
When I ran `samtools mpileup` for the Antarctic reads with the default quality cutoff of 13, I got 20 reads which included position 3037 but zero of them had C3037T.
- Thesis of Toxic substance (VEA) present then removed:
Compare with the curve for the 1981 Spanish Toxic Oil Syndrome, page 7 of this doc:
http://www.euro.who.int/document/e84423.pdf
It does not have the slow curve up-front: a toxic batch has a massive up-front effect, which then fizzles off. The Toxic Oil was identified years later.
Yes EVALI was cover-up for early COVID.
We note who insists that it's not, with a level of scrutiny/exploration/data much different from what is applied to WIV/Wuhan-CDC or even Fauxi/Sadsack/Baryc.
What's at play here? Directed Limited Hang-out?
Who died unexpectedly, when?
When did it spread?
- 2009 "Swine" H1N1 Flu is clearly high & out of sync/very early.
Was that a rehearsal, or also a leak? a first attempt *also with a pandemic declaration*, but that fizzled too fast, and also a deadly vakscene withdrawn after maiming a couple hundreds?
The next big ILI curve is 2017-2018 cyan, better synchronized - while it was NOT in Europe
Page 4 of https://www.ecdc.europa.eu/sites/default/files/documents/AER_for_2019_influenza-seasonal.pdf
2017 first invisible contamination wave.