Finding Bifido

Bioinformatics; noun plural but singular in construction bio·in·for·mat·ics \ˌbī-ō-in-fər-ˈma-tiks\

...the collection, classification, storage, and analysis of biochemical and biological information using computers especially as applied to molecular genetics and genomics (Merriam Webster).

Apologies to Disney and MicrobeWiki

Sometimes we want to know which tiny species of bacteria are inhabiting a particular ecosystem. Maybe we want to know what bacteria make us healthy, or maybe we want to know what bacteria just killed a dozen diners. Maybe we just want to see who is living in our bellybutton. Join me as I take you on a journey in "Finding Bifido."

There are several ways to discover microbial life. The easiest is by looking at a droplet containing microbes under a powerful microscope and comparing what you see to a picture, much like picking out a mugger in a lineup down at the 12th Precinct.

NatGeo

By looking at bacteria under a microscope, we can learn much about its identity, but it is nearly impossible to accurately identify a particular species of bacteria. Bacteria are rather elusive to the researcher armed with only a microscope, so they've developed ways of dying the bacteria which effectively divides all bacteria into two groups:  Gram positive and Gram negative.    

Microscopic image of gram-negative Pseudomonas aeruginosa bacteria (pink-red rods)

By examining how a bacteria reacts to dye as well as how it behaves, grows, and dies, researchers have done a great job finding bacteria for the last couple hundred years. Finding bacteria in this manner requires that the bacteria be "culturable," unfortunately, only a small percentage of bacteria are willing to be grown in a Petri dish. A scanning electron micrograph imager lets the researcher see an individual bacteria:

Wikipedia

In a spoonful of garden dirt, there can be trillions of bacteria. Even with the most expensive microscope in the world, you'd never be able to identify all of the little guys. So researchers have come up with a way to cheaply and quickly look at the genetic material found in bacteria to identify them.

DNA

Bacteria are living creatures. They have DNA, just like all lifeforms, and this DNA can give clues to exactly what species of bacteria we hold in our hand.

Up until the '90's, DNA was a simple concept. It was shown as a double-helix made up of bonding pairs of chemicals labeled ACT and G. We now know that DNA can take many shapes and be made of one, three, and even four strands.

Scientific American

We can solve crimes using the clues left behind by a tiny drop of DNA!

Berkeley.edu

 And determine paternity.

Daily Mail

But picking out a criminal or a baby-daddy is much easier than identifying each of the trillions of bacteria that may inhabit a single drop of swamp water. Here is where biology meets technology!

16s rRNA

DNA can be used to compare a couple of samples, but it generally contains simply too much information to be useful. RNA is a product of DNA. Sections of DNA ("genes"), produce copies of themselves called RNA. These RNA strands are single chains, not double-helix, but they are still made of the same chemicals we earlier labeled A, C, T, and G.

Note: At this point we could veer off into never-never land and discuss RNA until the end of time. Science is still learning about RNA and all its forms...don't ever let anyone fool you into believing that we know everything about DNA and RNA.  The "junk DNA" of the '80's is the "non-coding Piwi-interacting micro RNA" of 2015.

Ribosomal RNA (rRNA) is a special molecule found in the ribosomes of the cells of all living things. rRNA is an ancient structure, and as such, preserves the identity of its owner very well. A section of rRNA known affectionately as "16s" is a highly conserved region found in the rRNA of all lifeforms and is commonly used to quickly identify bacteria.

16s rRNA can be identified in each bacteria by it's strings of ACTG. Every 16s rRNA chain contains approximately 1500 letters. Laid out flat, it looks like this:

PNAS

   
And in 3D:

Wikipedia

If this is starting to seem immensely complex, it's not. To a computer geek and a microbiologist, this presents no problem at all!

Identifying Bacteria

The first step in finding 16s rRNA involves chopping the bacteria up into tiny little pieces to extract its DNA and RNA.  This chopping is done using standard chemicals found in biotech supply houses, and even on Amazon!

Chemicals used in this process:

1. TE buffer: It is a buffer used for the storage of nucleic acids (DNA and RNA), and also to prevent it from degradation.
2. Lysozyme: Enzyme that is used for degrading the cell wall of the organism.
3. SDS: SDS is strong anionic detergent that helps in solubilizing the proteins and lipids present in the membranes. It exposes the chromosomes that contain the DNA, also helps in releasing DNA from histones and other DNA binding proteins by denaturing them.
4. Proteinase K: Degrades most of the protein impurities in the DNA(Deproteination).
5. Phenol: Helps in removing most of the protein impurities from the DNA.
6. Chloroform: Prevents the shearing effect of DNA during the extraction process.
7. Isoamyl alcohol: Reduce the formation of the foams in extraction techniques.
8. Sodium acetate: Precipitates DNA in the solution
9. Absolute ethanol: Precipitate the DNA out of solution.

Don't worry!  No test at the end.

The next step in identifying bacteria from this mash-up of "gut bug guts" is known as Polymerase Chain Reaction (PCR) amplification. PCR amplification is simple, but would take too long to describe. It was invented in 1983, and garnered a Nobel Prize for its inventor in 1993. Here again, a couple of chemicals and a few pieces of lab equipment, and viola!

Every piece of DNA chopped up in the extraction process, and subjected to 30 cycles of PCR, produces about one billion copies of each strand. The "primers" used ensure only segments of 16s rRNA are selected. Easy peasy!

At the end of the PCR process, we have a vial filled with segments of 16s rRNA that are random chains roughly 150 "letters" long. The fluid containing all these random segments is sorted using various techniques involving electrical current and certain gels that will quickly stratify the chains of 16s rRNA into a format that can then be read by a computer. The final step looks like this:

OpenI

Bioinformatics

Biology meets information meets mathematics.

The field of bioinformatics has exploded in the past 20 years. What scientists spent years doing, aided by university supercomputers, can now be done at home in seconds on the crudest of home computers.

16s rRNA extraction culminates in the creation of a computer file. This file has a standard designation ".fasta". These fasta files are used widely in bioinformatics, but for the purposes of bacterial identification using 16s rRNA, they all look similar. If you open a fasta file, you'll see a ">" symbol followed by a whole bunch of A's, C's, T's and G's. These fasta files can be modifed to contain quality information (known as .fastq) or amino acid sequences (".faa"). 

Finding Bifido

 

Companies that perform bacterial analysis using 16s rRNA sequencing such as uBiome or American Gut Project, provide a detailed analysis of samples of skin, mouth, genital, or fecal samples that patrons provide (for a fee!). 

These companies will give a detailed report of what the sample contained, some companies also provide the fasta (or fastq) files associated with your report, called "raw data.". These reports can be a source of endless fun for the bioninformatician in you! 

Let's Find Bifido!

Once downloaded and unzipped, you'll find a standard fasta file. To open this file, you'll need to go to a free service like Galaxy.

When converted to a format you'll recognize, you'll see thousands and thousands of lines of ACTG's.

The start of each sequence is denoted by a ">", like this: 

>NS500457:47:H25CFAFXX:2:11101:10270:1118 1:N:0:GAGTGTGCCAGCAGCCGCGGTAATACGTAGGGTGCAATCGTTATCCGGAATTATTGGGCGTAAAGGGCTCGTAGGCGGTTCGTCGCGTCCGGTGTGAAAGTCCATCGCTTAACGGTGGCTCCGCGCCGGGTACGGGCGGGCTTGCGGGCGGT

The first stretch of numbers and letters (>NS500457:47:H25CFAFXX:2:11101:10270:1118 1:N:0:) is information about this sample and how it was processed. Computer programs understand this...you don't have to.

There are also free programs that will match this sequence against repositories of sequences that researchers all over the world have so kindly donated for the sake of science. Once such free program is Genome. Just copy the fasta sequence, choose a few options that denote RNA and which database you'd like to use, and click "Compute." In just a few seconds, you be provided a list of species that this sequence closely resembles.  The sequence above results in 500 BLAST "hits". The #1 most likely suspect, Bifido!:

This graphic shows that our sequence of approximately 150 letters (Query) matched almost perfectly from position 1 through 94 to a "Subject."  The subject's code is at least 618 letters long, and the match was in the region from 525 through 618.  By clicking on the file highlighted in the BLAST results page, we see that this is the full sequence we believe we have a match with:

>JQ190735.1.1349 Bacteria;Actinobacteria;Actinobacteria;Bifidobacteriales;
GAUGAACGCUGGCGGCGUGCUUAACACAUGCAAGUCGAACGGGAUCGGCUGGAGCUUGCUCCGGCCGUGAGAGUGGCGAA
CGGGUGAGUAAUGCGUGACCGACCUGCCCCAUACACCGGAAUAGCUCCUGGAAACGGGUGGUAAUGCCGGAUGCUCCAGU
UGGAUGCAUGUCCUUCUGGGAAAGAUUUAUCGGUAUGGGAUGGGGUCGCGUCCUAUCAGGUAGUCGGCGGGGUAACGGCC
CACCGAGCCUACGACGGGUAGCCGGCCUGAGAGGGCGACCGGCCACAUUGGGACUGAGAUACGGCCCAGACUCCUACGGG
AGGCAGCAGUGGGGAAUAUUGCACAAUGGGCGCAAGCCUGAUGCAGCGACGCCGCGUGCGGGAUGACGGCCUUCGGGUUG
UAAACCGCUUUUGACUGGGAGCAAGCCCUUCGGGGUGAGUGUACCUUUCGAAUAAGCACCGGCUAACUACGUGCCAGCAG
CCGCGGUAAUACGUAGGGUGCAAGCGUUAUCCGGAAUUAUUGGGCGUAAAGGGCUCGUAGGCGGUUCGUCGCGUCCGGUG
UGAAAGUCCAUCGCUUAACGGUGGAUCCGCGCCGGGUACGGGCGGGCUUGAGUGCGGUAGGGGAGACUGGAAUUCCCGGU
GUAACGGUGGAAUGUGUAGAUAUCGGGAAGAACACCAAUGGCGAAGGCAGGUCUCUGGGCCGUCACUGACGCUGAGGAGC
GAAAGCGUGGGGAGCGAACAGGAUUAGAUACCCUGGUAGUCCACGCCGUAAACGGUGGAUGCUGGAUGUGGGGACCAUUC
CACGGUCUCCGUGUCGGAGCCAACGCGUUAAGCAUCCCGCCUGGGGAGUACGGCCGCAAGGCUAAAACUCAAAGAAAUUG
ACGGGGGCCCGCACAAGCGGCGGAGCAUGCGGAUUAAUUCGAUGCAACGCGAAGAACCUUACCUGGGCUUGACAUGUUCC
CGACAGCCCCAGAGAUGGGGCCUCCCUUCGGGGCGGGUUCACAGGUGGUGCAUGGUCGUCGUCAGCUCGUGUCGUGAGAU
GUUGGGUUAAGUCCCGCAACGAGCGCAACCCUCGCCCUGUGUUGCCAGCACGUCAUGGUGGGAACUCACGGGGGACCGCC
GGGGUCAACUCGGAGGAAGGUGGGGAUGACGUCAGAUCAUCAUGCCCCUUACGUCCAGGGCUUCACGCAUGCUACAAUGG
CCGGUACAACGGGAUGCGACCUCGUGAGGGGGAGCGGAUCCCUUAAAACCGGUCUCAGUUCGGAUUGGAGUCUGCAACCC
GACUCCAUGAAGGCGGAGUCGCUAGUAAUCGCGGAUCAGCAACGCCGCGGUGAAUGCGUUCCCGGGCCU

Notice anything funny?

That's right!  The "T's" have all been replaced with "U's" in this file. Some researchers use a U instead of a T when discussing RNA. BLAST recognizes this and didn't miss a beat.

Another strange thing you may notice is that the #1 BLAST hit is to an "uncultured" Bifido species. If we want to drill in even deeper, we need more information!

Reverse fasta

RNA sequences always proceed in an orderly fashion, however, as they are produced in a cyclone of whirling chemical reactions (Fasta! Fasta!), no one can say if the end result of 150 letters is presented forward, backward, or even in mirror-image.  Some of the providers of these reports also give the raw data in a reverse fasta file. This reverse file can be identified by having the exact same "informational text" in the very first line, but differentiated slightly in some way to show it's a reverse of the first. This reverse file is not a simple reversing of the letter sequence, but a completely different analysis of the first sample, done in reverse, so it may result in a completely different outcome when BLASTed. Easier to just watch!

The reverse fasta of our "Finding Bifido" sequence looks like this (Obtained by uploading raw data fasta file into Galaxy):

 >NS500457:47:H25CFAFXX:2:11101:10270:1118 2:N:0:GACACCGGACTACTGGGGTTTCTAATCCTGTTCGCTCCCCACGCTTTCGCTCCTCAGGTCAGTGACGGCCCAGAGACATGCCTTCGCCATTGGAGTACTTCCCGATATCTACACATTCCACCGTTACACCGGGAATTCCAGTCTCCCCTA

The only difference on the informational line is the "2" (highlighted in pink). The sequence that follows the descriptor is a totally new set of ACTG, which will we want to BLAST, resulting in this #1 hit:

 

This is the exact same match that the first sequence got us, but do you notice anything different? The first sequence that we BLASTed, resulted in a sequence match from letters 1 through 94 of our "query" using letters 525 through 618 of the subject. In this sequence match, the matching letters were found in letters 618 through 711, and they are listed backwards!  This is a big clue that our second file is, indeed, a reverse fasta.

Reverse-complementing 

"My, you sure look ugly today!" Ooops, not that kind of reverse compliment?

In reverse-complementing, we need to look at the sequence in mirror image, and also switch the ACTGs to match their natural binding partners.  Sounds complicated, but it's not. There's an app for that!

Back in Galaxy, we find an option to "reverse-complement:" our sequence, the reverse complement of the above file looks like this:

>NS500457:47:H25CFAFXX:2:11101:10270:1118 2:N:0:TAGGGGAGACTGGAATTCCCGGTGTAACGGTGGAATGTGTAGATATCGGGAAGTACTCCAATGGCGAAGGCATGTCTCTGGGCCGTCACTGACGCTGAGGAGCGAAAGCGTGGGGAGCGAACAGGATTAGAAACCCCAGTAGTCCGGTGTC

In this version, everything in the informational line is the same (note the pink "2"), but all of the C's are now G's and the A's are now T's and the new sequence is backwards.  Confirm this by looking at the first 5 letters of the original sequence: " GACAC" and comparing them with their mirror image in the last 5 letters of the new sequence: "GTGTC."  Get it?  Simple!  Mirror imagery: GTGTC|GACAC

Don't worry, some people's brains don't see things like this...that's why there's an app.  Can you imagine doing that by hand?  

So what if we BLAST this reverse-complemented file?  We get this:

It's the same exact file we've been getting, still at #1, but now with a bit less confidence. The first 2 BLASTs were shown as 97% indentity, this one is 95%. Also some of the sequence numbers are a bit shorter. However, we now see in the "Subject" sequence, that the order has again been reversed, meaning that by "reverse-complementing" this file, we have lined it up with the original sequence. 

Joining Files

Here comes the fun part!  Until now, we've been dealing with sequences 150 letters long, and only using 80 or 90 of the letters in the #1 BLAST hits.  What if we simply join these two files?  Again, there are apps for that, but we can also easily do this manually by cutting and pasting one to the other, resulting in a 300 letter sequence:

GACACCGGACTACTGGGGTTTCTAATCCTGTTCGCTCCCCACGCTTTCGCTCCTCAGCGTCAGTGACGGCCCAGAGACATGCCTTCGCCATTGGAGTACTTCCCGATATCTACACATTCCACCGTTACACCGGGAATTCCAGTCTCCCCTATAGGGGAGACTGGAATTCCCGGTGTAACGGTGGAATGTGTAGATATCGGGAAGTACTCCAATGGCGAAGGCATGTCTCTGGGCCGTCACTGACGCTGAGGAGCGAAAGCGTGGGGAGCGAACAGGATTAGAAACCCCAGTAGTCCGGTGTC

And BLASTed:

Finally, we end up with a different result. This last BLAST, formed by reverse complementing the reverse file and joining it with the forward file, resulted in a 96% match with a different file, and this sequence shows that we are looking at the species Bifidobacteria adolescentis.  I believe we just "Found Bifido!"

Were the extra steps worth it to discover which species of Bifido we were looking at? I believe so. Some sequencial analyses are not so straight forward, result in top 5 hits of completely different families or phyla of bacteria. It just so happens that Bifidobacteria is very well studied and lots of researchers have entered their databases into public repositories for all to use.

In some regards, knowing the bacterial species may not be as important as knowing the genus or even phylum. In case you've forgotten, a small refresher in taxonomy:

Citylab

Conclusion

I hope you enjoyed this journey to Finding Bifido. The advances in science, biology, computers, and deciphering all this information is simply mind-boggling. As computers get faster and databases of bacteria grow larger, researchers will find new ways of determining exactly what bacteria call your body "home." For now, the science is still a bit young and requires a certain amount of guess-work, even when using computers.

For the advanced biohacker, the steps are like this (Thanks, Barney!):

Option 1 :
 
1. Combine all four R1 Lanes into a single file.
2. Combine all four R2 Lanes into a single file (Same lane order as step 1).
3. Reverse compliment the combined R2 file using a system like Galaxy or 
equivalent. If using Fastq format, handle the quality appropriately.
4. Join the sequences from Step 1 with the reverse complimented 
sequences from Step 3, sequence by sequence, to create a final file. 
Each sequence will have a length equal to the length of the R1 sequence 
+ the reverse complimented R2 sequence. Use the header from the R1 
sequence for the joined sequence.
 
Note : To allow a system like MG-rast to handle quality filtering, use 
Fastq format.
 
Option 2 :
 
1. Reverse compliment each of the four R2 lane files using a system like 
Galaxy or equivalent. If using Fastq format, handle the quality 
appropriately.
2. For each lane, join the sequences from the R1 file with the reverse 
complimented R2 file from Step 1, sequence by sequence, to create a 
final file. Each sequence will have a length equal to the length of the 
R1 sequence + the reverse complimented R2 sequence. Use the header from 
the R1 sequence for the joined sequence.
3. Combine the four lane files from Step 2 into a single file.
 
Note : To allow a system like MG-rast to handle quality filtering, use 
Fastq format.
 

Questions?

Later!

Tim