26 February 2021

SARS-CoV-2 was designed says Intelligent Design Theorist Michael Behe

Michael Behe (©Discovery Institute)
'exquisitely and purposively arranged'


Intelligent Design Theorist Michael Behe, author of Darwin's Black Box [2], was asked in the early days of the covid-19 pandemic to comment on the pandemic. He wrote a short blog.
"So, do I think viruses were designed? Yes, I most certainly do! The viruses of which we are aware — including the coronaviruses, Ebola, and HIV — are exquisitely, purposively arranged, which is the clear signature of intelligent design. [1].

Let's first try to understand what it means that viruses are designed and discuss later the evil designer. Behe tells us that viruses in general are designed including 'coronaviruses'. But coronaviruses consist of 4 genera with 21 species. Are they all designed? Behe does not go into details. The details are important because viruses differ in their potential to cause local outbreaks, epidemics or pandemics in animals or in humans. Behe tells us that they all are 'exquisitely, purposively arranged'. But 'purposively arranged' for what purpose? To infect bats or humans? How does he know? Should we conclude that SARS-CoV-2 was more 'exquisitely designed' than SARS-CoV-1 because the SARS-COV-2 caused a pandemic and SARS-CoV-1 only an epidemic? Apparently, there are unexplained differences in exquisiteness. The concepts 'exquisitely', 'purposively' are too vague to have scientific meaning. But Behe uses the concepts nonetheless without any specifications. If SARS-CoV-2 is so 'exquisitely and purposively' designed, would he be thrilled to be infected by this beautiful virus? [14].

And then there is a problem of timing of the origin. SARS-CoV-2 appeared in December 2019 in Wuhan, China. Was it designed at that time and place? To discuss this, we need a quote from Behe's article:

"Viral mutations can change the shape of the skeleton key at random. Most of the time the process doesn’t work but, every once in a while, the virus hits the jackpot."[1].

But this is a clear description of random mutation and natural selection! He continues: "In some cases, a viral key that had bound to a wild-animal protein mutates and becomes able to bind to a similar human protein" [1]. Here, he describes zoonosis, a jump from animal to human. But again, he describes it as a process of random mutation and natural selection. Winning the jackpot is luck, not design. Conclusion: SARS-CoV-2 originated by random mutation and natural selection. Behe does not notice the contradiction. 

It becomes even more interesting. SARS-CoV-2 has a unique 12 base / 4 amino acid insert at the contact site of the virus with the human receptor [6]. It is not present in other corona viruses. Since the beginning of the pandemic conspiracy theories spread on social media claiming that the virus was bio-engineered by China [7]. That is an intelligent design theory contradicting its natural origin. Behe ignores it. Why? It would be the perfect opportunity for ID theorists to apply their 'inference-to-design-methodology' to the origin of SARS-CoV-2 [8]. That method was designed to distinguish between natural and designed objects. Apart from Behe, William Dembski would be the right person to do it [9]. He could for once and for all give the definitive answer and silence all other conspiracy theorists. As far as I know no ID-er has done it so far. A missed opportunity. Why? Could it be that ID-ists already have an explanation: the virus was designed (by God). Could it be because ID-ists share the belief that SARS-CoV-2 doesn't have a natural origin? However, some ID-ists accept a natural origin [15].


Storm analogy

Behe: 

"...while of course the virus is dangerous, the situation can be compared to a strong storm on the ocean. The waves may be huge and the surface roiling, but the deeper waters continue as they always have, essentially undisturbed. In a similar way, although superficially it changes very rapidly, some researchers think that the coronavirus2 and many other virus types3 have remained basically the same for tens of millions of years" [1],[10].

Basically the same? SARS-CoV-2 stayed basically the same for millions of years and then suddenly caused a pandemic! When one particular virus "hits the jackpot" (Behe's words), how helpful it is to say it is basically the same? It does not help the scientific understanding of what causes a pandemic. I do not understand why Behe quotes two pre-pandemic publications [11],[12] if he wants to explain the origin of a new pandemic. The coronavirus group was created some tens of millions of years ago and did not change basically? So, there was no evolution? But the virus hit the jackpot! In his own words, the purpose of the storm metaphor is "most viruses do not affect humans and may well have a positive, necessary role to play in nature of which we are currently unaware". But even if true, is irrelevant and an inadequate respons. We are discussing SARS-CoV-2, and not other viruses. We are discussing a pandemic, not harmless viruses.

The most important question for virologists is: can we predict an outbreak that could cause a pandemic? What information do we need? How can we be prepared? Behe's reaction: the virus stays basically the same, and there are good viruses too! Both remarks are irrelevant. He should have payed attention to 'hitting the jackpot'. He has the right biochemical knowledge to do that.

The two articles Behe uses to support his basically the same claim are both pre-pandemic (his note 2 is dated 2013 [11] and his note 3 is dated 05 December 2018 [12].) So,  March 2020 he uses pre-pandemic publications to show that SARS-CoV-2 is basically the same!


Evil by Design

Continuing the first quote:

"Well, then does that mean the designer is evil and wants people to suffer? No, not necessarily. I’m a biochemist, not a philosopher. Nonetheless, I see no reason why a designer even of such things as viruses should be classified as bad on that basis alone." [1].

Behe introduces the topic 'evil' in a discussion of the SARS-CoV-2 pandemic. This is nothing less than the infamous theological problem of evil [13] which is an argument against the existence of a benevolent God. Behe does not tell us what additional facts are required in order to classify the Designer as evil. What could be added? Please note, that Behe refrains from saying: 'the Designer is morally good'. Why? Anyway, Behe did not solve the problem of evil.

Behe has problems with a Darwinian evolutionary process of random variation and natural selection, a mindless process of trial and error. However, he has no problem with an Intelligent Designer who exquisitely (!) and purposively  (!) designed pathogenic viruses and bacteria such as malaria, Ebola, HIV and the current SARS-CoV-2 virus. At the time he could not know that there are now 28 million SARS-CoV-2/covid-19 cases and more than half a million deaths in the US alone and 2,5 million deaths worldwide [3], but I am sure that he did not change his mind about the moral character of the designer. What could change his mind? [16]. Behe thinks that the person who designed SARS-CoV-2 is not necessarily evil. A bioterrorist is defined as someone who uses biological agents, such as pathogenic organisms, for terrorist purposes. A bioterrorist is not necessarily evil? Then what would make a bioterrorist evil?

In The Edge of Evolution (2007) he wrote about malaria:

"Malaria was intentionally designed. The molecular machinery with which the parasite invades red blood cells is an exquisitely purposeful arrangement of parts. (...) What sort of designer is that? What sort of "fine-tuning" leads to untold human misery? To countless mothers mourning countless children? Did a hateful, malign being make intelligent life in order to torture it? One who relishes cries of pain? Maybe. Maybe not." (p.237) [2]

Again, he prefers that an Intelligent Designer designed the most horrible pathogenic bacteria and viruses, to accepting they are the result of a blind, purposeless, random, natural process. Think about this: one cannot blame a virus, but one can and must blame a person. Especially, if that person is a God who is said to be merciful and compassionate and the foundation of morality. 

Was it really necessary to create 24 families of pathogenic viruses with more than 260 virus species that infect humans? [17]. Why that abundance?  In total, there are about 1,400 known species of human pathogens (including viruses, bacteria, fungi, protozoa and helminths) [18]. A little bit overdone. God wasn't satisfied with only 1 or 2 pathogenic species? It must have been great fun to create such overwhelming pathogenic biodiversity.


Have a nice day

Years ago I was worried by an unsettling statement of anthropologist Jonathan Marks:

"The scientist says: Science has explained many things about the universe. Your life has no meaning. Have a nice day." [5]

I tried to find a good answer. Yes, I know that 'the meaning of life' is outside the domain of science. But still it worried me. Many years passed by before it occurred to me that religion (Christianity) had no comforting truths to offer to a sensible and sensitive person. On the contrary: God created Malaria, Ebola, HIV, SARS-CoV-2. Have a nice day. I would rather accept reality as it is and try to find cures for those illnesses, then believing that those horrible diseases are created by a morally good God and leave it at that. Have a nice day!



Updates

5 March: Comments are welcome. Because of problems, I disabled moderation, but I cannot disable the google image test. If you still have problems at adding a comment, please mail the comment to:

6 March: replaced heading 'Evil' by: 'Evil by Design'

14 March: some editing, moved a passage, added and replaced a sentence, but the conclusion is the same. 

16 March: added paragraph with notes 17 and 18.

Notes

  1. Michale Behe Evolution, Design, and COVID-19, March 10, 2020.
  2. Darwin's Black Box review; The Edge of evolution review.
  3. worldometers (US) worldometers (world) 24 Feb 2021
  4. See my review of that book on my WDW website. 
  5. Jonathan Marks (2003) "What it means to be 98% chimpanzee. - Apes, people, and their genes.".  As a reply here I collected a list of 'comforting religious truths' (enough for a future blog!).
  6. I blogged about the PRRA insert here.
  7. Just type 'sars-cov-2 engineered' in the search field of youtube!!!
  8. Did Intelligent Design Just Miss Its Corona Moment?, American Scientist July 16, 2020
  9. Mathematician William Dembski (1999) Intelligent Design. The bridge between science and theology , invented the design inference (Specified complexity). See my review of that book.
  10. Behe's refers to two mainstream publications based on evolution by random mutation and natural selection! See below: [11], [12]
  11. Wertheim, J. O. et al. 2013. A case for the ancient origin of coronaviruses. Journal of Virology 87:7039-7045.
  12. Simmonds, P., Aiewsakun, P. and Katzourakis, A. 2019. Prisoners of war — host adaptation and its constraints on virus evolutionNature Reviews Microbiology 17:321-328. published 5 December 2018. That is in the year that the pandemic started! Remarkably, the article says: "...rates of sequence change that are orders of magnitude greater than those of the hosts they infect. They display evolution in real time as they acquire antiviral drug resistance, mediate persistent infection through escape from T and B cell immune system responses to infection or, at the experimental level, rapidly adapt to different cell culture conditions, new receptors and new hosts." So, Behe quotes an article that demonstrates evolution! This is an important article about archaeo-virology and paleo-virology. 
  13. See for 'the theological problem of evil' for example my review of Swinburne's Is There a God?  and God, Hitler and the Free Will Defense on my WDW website. Website of Stanford Encyclopedia of Philosophy: Problem of Evil.
  14. this sentence added on Sunday 28 Feb 2021
  15. Jonathan Wells admits that SARS-CoV-2 could have evolved, so was not intelligently designed. Michael Egnor writes: "intelligent design of the COVID-19 virus seems unlikely." Both contradict Michael Behe. [2 Mar 2021]
  16. "It is estimated that viral infections contribute to approximately 6.6% of global mortality." Would that change his mind? (source) Added 5 Mar 2021.
  17. See table with humans pathogenic viruses in wikipedia. See a list of all viruses (wikipedia). Added: 16 March 21.
  18. Microbiology by numbers, Nature Reviews Microbiology 12 Aug 2011. Added: 16 March 21.

23 February 2021

How to investigate the origin of SARS-CoV-2 yourself

The real pleasure of doing science is finding new things yourself. I show you in this blog how you can make discoveries yourself. A few days ago I blogged about the video of Dr. Wilson. The good thing about his approach is that he helps us to find out basic facts about the origin of SARS-CoV-2 ourselves. In his video he gives a few hints how to do it, but does not give the full details. Here he shows how you can enter bat viral sequences in the NCBI software:

NC_045512, MG772933, MG772934

and the fourth: RaTG13. These are the identifiers of human SARS-CoV-2 RNA viral sequence and of the evolutionarily closely related bat viral sequences. Everybody can compare them. You don't need to be a professional virologist. You don't need any training. It is free access. You don't need an account.  Just a few mouse clicks as shown in this video:

video 1 min 36 sec

What the video shows: start with this NCBI virus screen. The Virus field should be empty because we are using only Accessions. Accessions are identifiers of specific sequences.

enter them in the Accession field and Submit.

For a start, I want to focus in the Spike protein. So, click on the  [Protein (70)] tab. Sort the Protein column. Select the surface glycoprotein, spike protein, spike glycoprotein by hitting 4 check boxes:

4 Spike protein accessions are selected


After hitting the Align button the real magic happens. The result is an alignment of the 4 protein sequences of the Spike protein. Optionally, hit the Sequence ID button to get the standard SARS-COV-2 sequence on top. In the video I show how to find the famous PRRA sequence. The PRRA sequence is very important and it is said to have caused the pandemic outbreak of SARS-COV-2 in humans [1]. The PRRA sequence sits in the middle of the Spike protein. The Spike protein gets the virus into human cells. It's an evolutionary innovation.

I guessed that PRRA must be somewhere in the middle of the 1273 AA Spike sequence, between position 600 en 700.

and there it is: the famous PRRA insertion!
 

A little bit sliding to the left and right, and there it is: on position 682! PRRA ! Sensation! That is the pleasure of discovering things yourself: PRRA is present in the SARS-COV-2 Spike protein and not in the 3 most closely related bat Spike protein sequences. It is an insertion relative to the other sequences. The alignment tool made it perfectly clear. If scientists told you it is there, it would not impress you very much. But seeing is believing. The gap in the other 3 sequences is not a real gap. It is a virtual gap necessary to align the rest of the sequence. There can be no real gap in a RNA or DNA string.

We are talking here about an insertion of only 4 amino acids: P R R A or 3 x 4 = 12 bases. Is that really a big evolutionary problem? Is all the fuss about 4 amino acids?

Please don't overlook the fact that the SARS-COV-2 Spike protein can be aligned with bat sequences! That means they are evolutionary similar enough that the Alignment tool succeeds in aligning them. Yes, there are many differences, but they are all single amino acids substitutions. So, they do not cause a complete failure of the alignment. Dr. Wilson points out that we don't expect these single amino acid substitutions if those bat viruses were used to create the SARS-CoV-2 virus in the lab. In fact, the engineered virus should be nearly 100% identical to an existing bat virus, except for the PRRA site. But that is not what we see. It would be pointless to create hundreds of pointless mutations. Conclusion: SARS-CoV-2 virus is not engineered.

Possible reply: but the putative engineers didn't use those bat sequences, but others! My reply: but where are they? Nobody have found them. The fact that nobody has found bat viral sequences 99,9% identical to SARS-CoV-2 is a problem for both hypothesis, the natural and the unnatural. Evolutionary biologists did not find the missing link, and 'Intelligent Design theorists' did not come up with the sequences that were used to create SARS-CoV-2.

Finally, we are comparing sequences with the first-ever sequenced SARS-CoV-2 in Dec 2019 in Wuhan in a human. Not with the highly mutated sequences of today. The bat sequences are older. The hunt is for bat viral sequences more recent or more similar to SARS-CoV-2. This blog will certainly not be the final blog about the origin of SARS-CoV-2!

I recommend readers of this blog to view the video of Dr. Wilson again (8 min) to have a better understanding of his arguments for the hypothesis that the SARS-CoV-2 virus has a natural origin. It doesn't make sense to repeat all his arguments here.


Notes

  1. The sequence at Spike S1/S2 site enables cleavage by furin and phospho-regulation in SARS-CoV2 but not in SARS-CoV1 or MERS-CoV, Nature, 9 Oct 2020.

 

Previous blog

21 February 2021

In een week van winterse sneeuw naar bloeiende crocussen met bijen, hommels en zweefvliegen

crocus met zgn. Blinde bij Eristalis tenax
Sony macro 90mm. 20 feb 21.

 
Pollenia species (vlieg) met stuifmeel (20 feb 21)
Sony macro 90mm


Aardhommel. 21 feb 21.
Sony macro 90mm

Pyjama-zweefvlieg  - Episyrphus balteatus.
Sony macro 90mm 22 feb 21

 

Bladpootrandwants. 20 feb 21.
Sony macro 90mm

Bladpootrandwants overwintert als volwassen dier en kan dan ook gedurende het gehele jaar gevonden worden met een sterke piek in september en oktober. De plotseling oplopende temperaturen van de afgelopen dagen zullen hem uit zijn winterslaap gewekt hebben. Nooit eerder gezien! Het is een exoot die Nederland verovert heeft. Een groot insect: tot 2 cm!
Bladpootrandwants (detail). 21 feb 21.
Sony macro 90mm

 

Wat heeft de Schepper toch veel aandacht aan minutieuze details van de Bladpootrandwants besteed! Details, die een gewone sterveling nooit van zijn leven te zien zal krijgen! Klik op foto voor vergroting!

20 February 2021

No, the coronavirus still was NOT made in a lab

 

 

 Corona Update 20 February 2021





Frank Visser alerted me to an educational video about the origin of SARS-CoV-2.

No, the coronavirus still was NOT made in a lab

The video is from Dr. Wilson (molecular biologist) and his channel is Debunk the Funk with Dr. Wilson. In this video he debunks the claim that SARS-COV-2 has been created in the lab. I haven't seen such a clear presentation. This is very welcome. Below the video are links to viral sequences and publications. He is not a virologist, but you can check out his claims with the help of the links he gives. There are a lot more videos in his channel. Recommended. 

 

17 February 2021

Klimaatontkennende Zwartkoppen en andere vogels

Gedurende de afgelopen vorstperiode van een week hadden we drie unieke vogelsoorten op bezoek die we nooit eerder in onze tuin hebben gezien: koperwiek, zanglijster en zwartkop.

vrouwtje zwartkop is gek op appel

Zwartkop en zanglijster kennen we heel goed, maar die hebben we nog nooit in onze tuin gezien. En dan ook nog in de winter! Normaal trekken alle zwartkoppen in de winter naar het Middellandse Zeegebied. Of verder. Deze niet. Dit vrouwtje behoort tot de kleine groep van klimaatontkenners! Ze negeren gewoon de hele winter en blijven hier. Dat scheelt een paar duizend kilometer heen en weer vliegen! Je moet dan wel bij mensen om voedsel bedelen, maar die vinden dat niet erg. Ik heb eens opgezocht hoeveel zwartkopklimaatontkenners er in Nederland zijn. In januari dit jaar waren dat 171 gevalideerde waarnemingen. Dat zijn waarnemingen op zicht, want in die maanden wordt er niet gezongen. Dat wordt weer gecompenseerd doordat ze in tuinen makkelijker waar te nemen zijn dan in het bos. Context: in april 2020 werd  er een record aantal van een kleine 10.000 gevalideerde zwartkop waarnemingen geregistreerd. In april zingt de zwartkop volop. En daardoor is hij (we nemen aan dat het een hij is) op afstand te herkennen. Voorzichtig zou je kunnen stellen dat er een trend is vanaf 2004 dat er meer zwartkoppen in Nederland overwinteren. Maar de data over die termijn bevatten een aantal onzekerheden.

Aantallen gevalideerde waarnemingen van zwartkoppen per maand in 2020.
Maximum in april, maar niet nul in dec, jan, feb.
Het gaat om de Waarnemingen en Individuen.
waarneming.nl

Het aantal overwinterende zwartkoppen is dus een relatief klein aantal. Maar, zou die groep een ondersoort in wording kunnen zijn? Dat is mogelijk als ze zich niet mengen met de groep van de echte trekvogels. En dat kan makkelijk, want ze zijn in de winter onder elkaar. Ze kunnen beginnen met paarvorming en broedterritorium afbakenen in maart zonder last te hebben van de anderen. Als het gemis aan trekinstinct erfelijk is, overwinteren de jongen ook in Nederland en -nog belangrijker- paren binnen de groep van niet-trekkers. Je hebt dan het begin van reproductieve isolatie (reproductive isolation). En dat is een voorwaarde voor soortvorming (speciation). Het begint als ondersoort. Maar, de reproductieve isolatie zal niet compleet zijn, want ze kunnen later in het seizoen nog een gemengd paar vormen. Tenminste als er geen uiterlijke kenmerken zijn waaraan de groepen elkaar zouden kunnen herkennen.

Over ondersoorten gesproken. En dit is voor de liefhebbers. In waarneming.nl tref je twee zwartkoppen aan: Sylvia atricapilla en de ondersoort Sylvia atricapilla atricapilla. Alles wat ik hierboven heb geschreven gaat over Sylvia atricapilla. Ik heb geen idee wat het verschil is tussen die twee en hoe mensen die ondersoort menen te herkennen. Waarneming.nl geeft geen hulp, geen details. Maar er zijn een klein aantal gevalideerde waarnemingen van de ondersoort. Soms met een foto of geluidsopname. Willekeur? Foutje? Ambtenarij? Geheimzinnig. Dit vereist nader onderzoek! [1]. Leuke vraag: kent ObsIdentify het verschil tussen soort en ondersoort?

De dorstige zanglijster: een stevige drinker!



tweede versie met muziek van Taede Smedes (4 maart 21)

zeer dorstige zanglijster (video 1min 30sec)

Zanglijster 16 februari 2021

Bij mijn youtube filmpje van de zanglijster heb ik al een commentaar geschreven. Ik denk dat het stevige drinken 's ochtends vroeg veroorzaakt wordt door: (1) lage luchtvochtigheid waardoor het beestje 's nachts uitdroogt. Hij kan nu eenmaal niet stoppen met ademhalen. En dat geeft vocht verlies. (2) er is door de aanhoudende vorst nergens stromend water te vinden. (3) water is toch iets gebruikersvriendlelijker dan sneeuw happen, dus drinkt hij in tuinen waar vers water aanwezig is. Ook ik zelf heb 's ochtends bij het wakker worden dorst. Dus ik snap het wel!


Koperwiek: een buitenlandse gast


Koperwiek 10 februari 2021

Koperwiek eet klimop bessen 11 februari 2021

De koperwiek is fraaie wintergast uit Scandinavië. Hij behoort tot het geslacht Turdus net als onze merel en zanglijster. Hij eet bessen van de klimop. Ik heb hem niet gezien bij het voer dat wij (geheel gratis) neergelegd hebben. Hij laat zich wegjagen door merels:

Merel man nadert koperwiek...

Koperwiek vertrekt geheel vrijwillig.
Hij begrijpt de suggestie: opdonderen!
Jij hoort hier niet! Dit is mijn tuin!

Noten

  1. Individual variability and versatility in an eco-evolutionary model of avian migration is onderzoek naar migratie van de zwartkop Sylvia atricapilla. Zeer interessant!

12 February 2021

Viruses are a textbook example of evolution by natural selection


Darwin & Corona Update 12 February 2021


If there is any Evolution textbook published long before the SARS-CoV-2 pandemic that makes the importance of viruses for evolution and human society clear right from the start, it is Freeman and Herron (2006). Chapter 1 is entitled: 'A Case for Evolutionary Thinking: Understanding HIV'. HIV is a RNA virus, just as SARS. The chapter explains the origin, evolution, transmission, treatment, natural resistance, and vaccination. But, there is more. Chapter 14: 'Evolution and Human Health' discusses Flu virus evolution, the evolution of antibiotic resistance, and how virulence evolves. 

In future editions of the textbook SARS-CoV-2 will undoubtedly replace HIV because it is researched in such unprecedented detail.  

 

Mendels Demon by Mark Ridley was published 20 years ago, but his discussion of the relation between virus mutation rates and genome size could not be more relevant today in the midst of the SARS-CoV-2 pandemic. I added a corona update highlighting his prescient insights: Mendels Demon.


 

 

Carl T. Bergstrom, Lee Alan Dugatkin (2011) 'Evolution': SARS coronavirus is present with schematic genome layout, and two kinds of overlapping genetic code is nicely illustrated (page 342–343). Several pages on viruses in general. See further: Introduction.
 

 

 

Douglas J. Emlen, Carl Zimmer 'Evolution. Making Sense of Life' (second edition 2013): SARS and corona virussen are discussed. An evolutionary tree of the corona viruses is shown in chapter 18 'Evolutionary Medicine', page 609. The data of bat, palm civet, and human viruses are based on the SARS outbreak in 2003. Obviously, the current SARS-CoV-2/covid-19 pandemic is not covered, but much of the discussion in that chapter is relevant for understanding the current pandemic. In the first chapter part 1.2 Viruses: The deadly escape artists the 2009 H1N1 Mexican flu pandemic is well explained with good color illustrations (pp.16-22). Please note a third edition is published in 2019, but I don't know if there is a big difference. 

 

--- new book ---

 


'A Most Interesting Problem: What Darwin's Descent of Man Got Right and Wrong about Human Evolution'

Edited by Jeremy DeSilva.

Princeton University Press

150 years ago, in 1871, Charles Darwin published The Descent of Man, in which he attempted to explain human evolution, a topic he called "the highest and most interesting problem for the naturalist." A Most Interesting Problem brings together twelve world-class scholars and science communicators to investigate what Darwin got right—and what he got wrong—about the origin, history, and biological variation of humans.
The good thing about this book is that the authors have no problem with recognising that some ideas of Darwin were wrong. Darwin is not infallible, only the Pope is
infallible.

Charles Darwin (12 Feb 1809 – 19 April 1882 )

08 February 2021

Verbeterde versie ObsIdentify herkent meer soorten, maar maakt het ook minder fouten? Deepfake portretten getest

"Verbetering beeldherkenning.
Deze week gaat onze nieuwe versie van de beeldherkenning live. De herkenning van het aantal soorten wordt opgeschroefd naar 22.303 (was 16.148)! De resultaten zijn aanzienlijk beter maar de AI is ook voorzichtiger (vooral bij dag- en nachtvlinders)." [1]

Dit stond eind december 2020 op waarneming.nl.

Mijn vraag: als ObsIdentify beeldherkenning beter is geworden in het herkennen van planten en dieren, is het dan ook beter in het onderscheiden van dieren en mensen? Ja, ik weet het: ObsIdentify is nooit getraind met afbeeldingen van mensen. Duidelijk. Maar als het beter is geworden in het onderscheiden van buizerd en ruigpootbuizerd, moet je dan niet verwachten dat het ook beter is geworden in de veel gemakkelijker opgave om mens/dier te onderscheiden? Uit pure nieuwsgierigheid naar de werking van beeldherkenning heb ik dit uitgeprobeerd.

Voor dit blog heb ik 514 deepfake portretten aan ObsIdentify aangeboden. De meerderheid gaf een percentage lager dan 40% en een aantal lag tussen de 40% en 90%. Verrassing: er waren portretten die 90% en hoger scoorden. Hier de winnaars:

ObsIdentify voorspelt Langpootmug (familie) onbekend - Tipulidae indet. met zekerheid 91.4%

ObsIdentify voorspelt Gewone heremietkreeft - Pagurus bernhardus met zekerheid 95.0%

ObsIdentify voorspelt Euraziatische Rode Eekhoorn - Sciurus vulgaris met zekerheid 96.1%

ObsIdentify voorspelt Roze Pelikaan - Pelecanus onocrotalus met zekerheid 97.6%

ObsIdentify voorspelt Roze Pelikaan - Pelecanus onocrotalus met zekerheid 99.9%

De winnaar:
ObsIdentify voorspelt Roze Pelikaan - Pelecanus onocrotalus met zekerheid 100.0%

In totaal scoorden 9 afbeeldingen hoger dan 90%. Eén afbeelding scoorde zelfs 100%. Let op: de foto's zijn op geen enkele wijze gemanipuleerd. Zo vers uit de computer. Alle plaatjes zijn 1024x1024 pixels. Ik heb ook geen andere afmetingen of uitsneden geprobeerd. Zo ziet dat er uit in waarneming.nl vóórdat je op de Accepteer button hebt geklikt:

100% Roze Pelikaan. (screenshot 6 feb 2021)

In feite zou ik op de Accepteer button kunnen drukken en vanwege de 100% zou het algoritme van waarneming.nl het plaatje automatisch goedkeuren zonder dat er een menselijke moderator aan te pas komt [2]. De onzekere voorspellingen zijn altijd vermakelijk. Bij de laatste foto: Ooievaar, Chinese knobbelgans en Lepelaar. Dus: vogels in de nabijheid van water? Voor mij een raadsel waarom er bij 100% ook nog onzekere voorspellingen zijn. 100%-zeker is niet 100% zeker? Dit alles werpt een nieuw licht op het concept 100% zekerheid.

Conclusie: Ondanks verbeteringen maakt ObsIdentify nog steeds hetzelfde type fouten met afbeeldingen buiten haar eigen domein. Zie vorige blogs. Het begint er dus op te lijken dat dit een permanente karakteristiek is van de software.

De Deepfake afbeeldingen zijn afkomstig van:

  https://www.thispersondoesnotexist.com/

Dit type software heet Generative Adversarial Network: het creëert beelden in plaats van ze te analyseren zoals ObsIdentify doet. Buitengewoon fascinerend omdat ze zo realistisch zijn. Zoals de naam van de website al suggereert: deze personen bestaan niet. ObsIdentify denkt daar anders over: het zijn duidelijk Roze Pelikaan, Euraziatische Rode Eekhoorn en Gewone heremietkreeft. Overigens zie je dat het programma dat gezichten genereert ook rare fouten maakt. Maar dat is een ander verhaal. De complete serie van 11 foto's en met de beoordeling van ObsIdentify is hier te vinden.

In een volgend blog ga ik plaatjes die ik eerder getest heb met de oude versie van ObsIdentify vergelijken met de nieuwe versie.

 

Noten

  1. Het bericht is na enkele dagen spoorloos verdwenen. Nergens op de website is het versie nummer van ObsIdentify te vinden. De versie moet dezelfde zijn als op de smartphone: ObsIdentify 1.4.2.
  2. Het toeval wil dat alle Roze Pelikaan waarnemingen in Nederland van 2020 het kenmerk "(nog) niet te beoordelen" hebben. Ik weet niet of dat er iets mee te maken heeft. Het lijkt mij dat dat geen invloed kan hebben op het functioneren van ObsIdentify op zich.

 

Vorig blogs over ObsIdentify:


  1. 30 September 2019 Test van ObsIdentify algoritme voor automatische identificatie van dieren en planten deel 1
  2. 21 Oktober 2019 ObsIdentify software gekraakt! Vlinders identificeren zonder te weten wat een vlinder is... deel 2
  3. 19 November 2019 ObsIdentify (3) Who is afraid of red, yellow and blue? De Vlinder Turing test voor mens en AI deel 3
  4.  4 December 2019 ObsIdentify herkent Kuifeend, Kikker, Kiekendief en vele andere soorten in plaatjes van 1 pixel deel 4
  5. 23 december 2019 Hoe zeker is 100%? Soorten met 100% zekerheid herkennen in random pixels. deel 5
  6.  4 Feb 2020 Hacken voor dummies en gevorderden. Beeldherkenningssoftware ObsIdentify is makkelijk te misleiden. deel 6 
  7. 18 maart 2020  ObsIdentify geeft ALTIJD foute antwoorden buiten zijn eigen domein. Ook met hoge zekerheden. deel 7.
  8. 30 mei 2020 Obsidentify voorspelt Wespendief, Buizerd en Ruigpootbuizerd op basis van foto's van dezelfde vogel. deel 8. 
  9. 26 Jun 2020: Toch nog een wespendief! deel 9
  10. 22 juli 2020: Tesla beeldherkenning en ObsIdentify beeldherkenning: steeds beter, maar maken nog steeds klassieke fouten deel 10.
  11. 11 Jan 2021: ObsIdentify herkent Cetti's zanger op tegenlicht foto met 99% zekerheid deel 11

07 February 2021

Sneeuw. De vogels komen naar je toe! Vink, koperwiek, putter, zwartkop, zanglijster, mus

 

vrouwtje vink. 7 februari 2021

Koperwiek. 7 februari 2021

Koperwiek, 7 februari 2021

mannetje vink. 8 feb 2021

Putters. 8 feb 2021

vrouw zwartkop. 8 feb 2021

Het moet toch niet gekker worden! De zwartkop is een trekvogel en overwintert in Zuid-Engeland, Spanje, Marokko en Algerije. Het is een zangvogel en insecten-eter. Hier zit ze bessen van de klimop te eten. Ik vond meerdere winterwaarnemingen in Nederland! Nieuw voor mij.

Zanglijster. 10 feb 2021

Zanglijster lijkt wel een beetje op de koperwiek, maar heeft geen rode flanken en geen duidelijke lichte oogstreep. Ook lijkt de zanglijster een beetje op een vrouwtje merel, maar de zanglijster heeft een veel lichtere borst.

vrouwtje huismus 11 feb 21

 

03 February 2021

New feature in NCBI virus database: View Mutations in SARS-CoV-2

 

 

 Corona Update 3 February 2021




There seems to be a competition between countries to report new SARS-COV-2 variants. The media try to make sense of it and try to answer questions about how dangerous these new variants are. For example, the Scientific American:  The Most Worrying Mutations in Five Emerging Coronavirus Variants [1] and The Scientist [5].

This is a very useful article. I will return to it. But there are more variants and many more mutations. What is the total number of different mutations that have been found worldwide up to now? Answer: NCBI virus database [2]. The NCBI started an overview of all mutations in SARS-CoV-2. This is free information and no account is required. This is a user-friendly website. 

View Mutations in SARS-CoV-2 SRA Data

Click on the link View Mutations:

Table with all mutations of SARS-CoV-2

After a few seconds a table with all mutations appears with columns. See appendix for the columns in the  list.

Explanation

A non-synonymous substitution is for example: D 614 G is : amino acid D is replaced by G in position 614 in the Spike (surface glycoprotein). The 614 position is relative to the start of the first amino acid (AA) of the protein. For the Spike protein the position is between 1 and 1273. That is the length of the protein. The Spike is a relatively small protein. 

The genomic position is a number between 1 and 29,903. That is the length of the standard reference SARS-CoV-2 genome.

A synonymous substitution for example: Q 613 QQ 'replaced' by Q. This is still a substitution because the substitution is at the nucleotide level: CAA > CAG. The nucleotide change is listed also in the table.

The Count gives an indication whether the mutation is rare. In Collected location the countries of origin of the virus sample are specified. 

Furthermore, a handy feature is that each column can be sorted (up/down) by clicking on the header. Try it!

There are not yet statistics provided by the NCBI website. I counted (30 Jan) the number of  mutations in Spike protein (surface glycoprotein):

  • 264 non-synonymous mutations
  • 345 synonymous mutations 
  • 609 mutations total

This is expected: there are more synonymous than non-synonymous mutations. This is quite a lot for a protein of 1273 Amino Acids: 20% Amino Acid changes and 47% of the Spike nucleotides have mutations. The million dollar question is what the effect is on the behaviour of the protein and the properties of the virus. A first step is:


From one-dimensional RNA to three-dimensional proteins

A spectacular and sophisticated feature is the interactive 3-D display of the protein which is shown when clicking on the link of the Protein Change. Try it!

Click on the link N501Y
 

Loading data ... please wait ... (ignore error message):

Interactive 3D model of Spike protein
mouse pointer at N501.

 


try full screen video! (16 sec)

By moving the mouse pointer over the protein, the names of individual Amino Acids with position are displayed. The software is keeping track of all 1273 Amino Acids in this very complicated 3D structure! Really great software! After a lot of trial and error I found the ASN501. 

ASN = Asparagin; 1-letter code: N. 
Tip: for the table of code names for amino acids see this page

Asparagin on position 501 (N501) is the location of the famous mutation N501Y. N is replaced by Y. The amino acid it is marked by a yellow color:

zoomed in. Yellow structure is Asparagin in position 501

Not surprisingly, the yellow position 501 is located on the outside of the molecule. It must attach to the human ACE2 receptor. It could not work if it were located at the inside of the molecule.

Try it. Play with it. Move the cursor over the structure. Manipulate the point of view with your mouse by holding the mouse button down and move. Watch the different angles of view. Try other mutations. (click on other mutations in the main table). Zoom in.  Mind you: this is the molecule that caused a pandemic!

Remember: the three-dimensional structure of a protein is the first step in discovering the effect of a mutation. 

Problems: Not all links to 3D proteins seem correct. H1000Q results in a protein THR257. The links are made manual?

Later I discovered that one can select certain locations in the one-dimensional RNA (in the right panel of the page) and the selected amino acid will appear yellow highlighted in the 3D model. I have to explore that.

 

The famous N501Y mutation is found in the variant in UK, South Africa and Brazil. Here is the list of the Scientific American article [1]:


  • Spain:        A222V (Spike)     -
  • UK:             -     -       N501Y   (Spike)
  • South Africa: E484K  K417N    N501Y [virus escape mutant]
  • Brazil:       E484K  K417N/ N501Y

 

Universe too small ! too short living !

The number of possible proteins of length 1273 is staggering. Do the calculation: for every position there are 20 possibilities because there are 20 Amino Acids. "So there are 20×20 = 400 distinct proteins of 2 Amino Acids, 20x20x20 = 8000 proteins of length 3 AA, 160,000 proteins of length 4 AA, 3,200,000 with just 5 AA." [4] etc. Total: 20^1273 AA sequences for the Spike. And that is only one protein! Obviously, evolution could not have tried out all those possibilities. The age of the universe is too short to try them all out! So, we can expect endless new virus variants coming as long as we don't interfere with the pandemic and the virus is allowed its natural course.


 Notes

  1. The Most Worrying Mutations in Five Emerging Coronavirus Variants, Scientific American,

 

Appendix

Information in the NCBI table with all mutations:

  • Protein: all proteins encoded by SARS-CoV-2
  • Amino Acid substitution (as far as I can see: no insertions/deletions...) 
  • Count: total number of cases in the database of the specific mutation
  • Genomic location: the position in bases or: nt
  • Codon change. For example: GCT > GCC  (T is used instead of U !)
  • Non-synonymous (does change AA) or synonymous (does not change AA), AA = amino acid.
  • Collection location: country of origin of the sample 

 

This page has a table with the abbreviations of the amino acids.

The video was created with SimpleScreenRecorder for Linux by Maarten Baert.