HC genomica deel 2
HC: BLAST
LUCA = last universal common ancestor. We believe that every life form existing on this
planet is developed from LUCA and that LUCA has given rise to all different species.
Bioinformatics: study of informatic processes in biotic systems.
Bioinformatic data analysis: using computational methods to analyze biological data.
Types of -omics:
- Genomics: sequence all of the DNA of one organism
- Transcriptomics: sequence all of the RNA in an organism/tissue/cell
- Proteomics: sequence all of the proteins in an organism/tissue/cell
- Metagenomics: sequence the DNA of all organisms in a sample
- Metatranscriptomics: sequence the RNA of all organisms in a sample
- Metaproteomics: sequence the proteins of all organisms in a sample
We use metagenomics for researching organisms because many organisms cannot be
grown in the lab or studied in the lab (only by their sequence) and because the
environmental conditions affect the presence and composition of organisms. For many
researches we filter the bigger organisms (like eukaryotes) out of the sample, so that we
only keep the prokaryotes. Afterwards we can isolate the DNA of the sample and
sequence this DNA by shotgun sequencing and with these short sequences we try to
figure out which organisms are present in the sample. → if you look at the prokaryotes
separately, the information about the sample (interactions) will not be the same as if we
look at all the prokaryotes together in metagenomics.
With the omics we have created the ability to sequence an entire genome of an
organism.
The omics revolution represents the transformative shift in life sciences from studying
single genes to comprehensively analyzing entire sets of biological molecules.
An existing bias in science: the organisms that have been sequenced the most are
associated with humans, because humans are mostly interested in themselves and
their needs. This causes biases in our general understanding of biology and in our
databases of sequenced organisms, because we look at a really small part of the
biodiversity on earth. With metagenomics however we can sequence all of it and create
a less biased look on biology. This however creates a lot of data, bioinformaticians use
data in 2 different ways:
- Question first (top-down): a biological question of hypothesis will lead to
datasets to answer this question/hypothesis.
, - Data first (bottom-up): a given dataset will lead to biological hypotheses or
questions that could help to answer/test the dataset.
After you have sequenced a genome or DNA sequence (with next generation
sequencing) you want to find another sequence (query → sequence that you are looking
for in a database) in a database that is similar to yours. If sequences are similar, they are
most likely related (they have the same function or belong to the same organism).
Evolution however can change small bits of a sequence, so to find a similar query to your
sequence you have to break your sequence down in k-mers (subsequences): consist of
k nucleotides or amino acids.
➔ Imagine a sequence of 7 nucleotides (ACCTGTA) and we want k-mers of 5
nucleotides. Then we make the following k-mers: ACCTG, CCTGT, CTGTA.
With these we can make an index of all k-mers that occur in the database sequences.
Due to evolution however, finding exact k-mers matches (if you only look at k-mers) is
limited. To prevent this limitation we do natural sequence divergence: we take our
database as the refence genome and compare this to multiple ‘similar’
genomes/samples. If sequences are closely
related (> 97%) to the refence genome
(probably from the same organism), they take
place at the top. If sequences are more
distantly related (60-80%) to the refence genome (might be from an organism that is
related to the one of our genome), they take place at the bottom. → gives us an idea of
where to place an organism in the tree of life.
Another way for finding a query for your sequence is pairwise sequence alignments: we
take 2 sequences and place these on top of each other and try to make sure that they
match as much as possible. → are really time consuming.
For the most optimal way of finding a query we combine the k-mers and pairwise
sequence alignments: we first find potential hits for a query using exact k-mers stored in
an index, and then we use pairwise sequence alignment only for the potential hits →
BLAST.
BLAST (Basic Local Alignment Search Tool) finds similar sequences/hits at a reasonable
speed. Heurtistics: a method that ‘cuts corners’ and therefore is much faster but can
also give not the most optimal result. → used by the fastest algorithms. The BLAST
search algorithm:
- Identifies all k-mers (length W) in the query
- Chooses the k-mers with the highest value (less common k-
mers are more infomative) → based on substitution matrix
, - Finds similar k-mers in the database (to exclude evolutionary influences) →
everything that scores above 12 in the substitution matrix with the original k-mer
is similar and involved in the BLAST.
- Extends seeds in both directions to find HSPs (regions that can be aligned with a
score above a certain threshold) between the query and a hit. → pairwise
sequence alignment.
➔ To only find closely related hits you make your k-mer
length longer and the neighbor threshold higher.
High scoring pair: part of the query and sequence where
the alignment score only goes up, and therefore we can say
that in this area our sequences can be related.
Local alignment: finds the optimal sub-alignment within 2 sequences. These are mostly
partial homologs (that part of the sequence has a common ancestor). → BLAST
exclusively makes local alignments.
Global alignment: aligns 2 sequences from end to end. If you know the 3 sequences are
full homologs (belong to the same organism).
BLAST input: in FASTA-format, query sequences. BLAST output: hits for the query
sequences.
BLAST forms:
- Nucleotide-nucleotide searches: blastn (finds
homologous genes in different species), megablast
(longer k-mer lengths, finds longer alignments
between very similar nucleotide sequences, best
tool for finding highly identical hits), discontiguous megablast (uses
discontiguous k-mers (excludes the 3d wobble base of an amino acid), focusses
the search on codons, best tool for finding nucleotide-nucleotide hits at larger
evolutionary distances).
- Protein-protein searches: blastp (finds homologous proteins in different species).
- Translated searches: blastx (put in nucleotides and BLAST translates your
sequences into a protein sequence, afterwards looks into protein database for
your translated sequence), tblastn (translates your protein sequence to
nucleotide sequence and look in the nucleotide database for your sequence),
tblastx (translates 2 nucleotide sequences to protein sequences and compares
the protein sequences, we do this because protein sequences are longer
conserved than nucleotide sequences).
➔ BLAST order: mask low-complexity regions – make a list of high scoring k-mers –
make a list of neighboring k-mers – make an index search with high scoring k-
mers – extend the alignment.
HC: BLAST
LUCA = last universal common ancestor. We believe that every life form existing on this
planet is developed from LUCA and that LUCA has given rise to all different species.
Bioinformatics: study of informatic processes in biotic systems.
Bioinformatic data analysis: using computational methods to analyze biological data.
Types of -omics:
- Genomics: sequence all of the DNA of one organism
- Transcriptomics: sequence all of the RNA in an organism/tissue/cell
- Proteomics: sequence all of the proteins in an organism/tissue/cell
- Metagenomics: sequence the DNA of all organisms in a sample
- Metatranscriptomics: sequence the RNA of all organisms in a sample
- Metaproteomics: sequence the proteins of all organisms in a sample
We use metagenomics for researching organisms because many organisms cannot be
grown in the lab or studied in the lab (only by their sequence) and because the
environmental conditions affect the presence and composition of organisms. For many
researches we filter the bigger organisms (like eukaryotes) out of the sample, so that we
only keep the prokaryotes. Afterwards we can isolate the DNA of the sample and
sequence this DNA by shotgun sequencing and with these short sequences we try to
figure out which organisms are present in the sample. → if you look at the prokaryotes
separately, the information about the sample (interactions) will not be the same as if we
look at all the prokaryotes together in metagenomics.
With the omics we have created the ability to sequence an entire genome of an
organism.
The omics revolution represents the transformative shift in life sciences from studying
single genes to comprehensively analyzing entire sets of biological molecules.
An existing bias in science: the organisms that have been sequenced the most are
associated with humans, because humans are mostly interested in themselves and
their needs. This causes biases in our general understanding of biology and in our
databases of sequenced organisms, because we look at a really small part of the
biodiversity on earth. With metagenomics however we can sequence all of it and create
a less biased look on biology. This however creates a lot of data, bioinformaticians use
data in 2 different ways:
- Question first (top-down): a biological question of hypothesis will lead to
datasets to answer this question/hypothesis.
, - Data first (bottom-up): a given dataset will lead to biological hypotheses or
questions that could help to answer/test the dataset.
After you have sequenced a genome or DNA sequence (with next generation
sequencing) you want to find another sequence (query → sequence that you are looking
for in a database) in a database that is similar to yours. If sequences are similar, they are
most likely related (they have the same function or belong to the same organism).
Evolution however can change small bits of a sequence, so to find a similar query to your
sequence you have to break your sequence down in k-mers (subsequences): consist of
k nucleotides or amino acids.
➔ Imagine a sequence of 7 nucleotides (ACCTGTA) and we want k-mers of 5
nucleotides. Then we make the following k-mers: ACCTG, CCTGT, CTGTA.
With these we can make an index of all k-mers that occur in the database sequences.
Due to evolution however, finding exact k-mers matches (if you only look at k-mers) is
limited. To prevent this limitation we do natural sequence divergence: we take our
database as the refence genome and compare this to multiple ‘similar’
genomes/samples. If sequences are closely
related (> 97%) to the refence genome
(probably from the same organism), they take
place at the top. If sequences are more
distantly related (60-80%) to the refence genome (might be from an organism that is
related to the one of our genome), they take place at the bottom. → gives us an idea of
where to place an organism in the tree of life.
Another way for finding a query for your sequence is pairwise sequence alignments: we
take 2 sequences and place these on top of each other and try to make sure that they
match as much as possible. → are really time consuming.
For the most optimal way of finding a query we combine the k-mers and pairwise
sequence alignments: we first find potential hits for a query using exact k-mers stored in
an index, and then we use pairwise sequence alignment only for the potential hits →
BLAST.
BLAST (Basic Local Alignment Search Tool) finds similar sequences/hits at a reasonable
speed. Heurtistics: a method that ‘cuts corners’ and therefore is much faster but can
also give not the most optimal result. → used by the fastest algorithms. The BLAST
search algorithm:
- Identifies all k-mers (length W) in the query
- Chooses the k-mers with the highest value (less common k-
mers are more infomative) → based on substitution matrix
, - Finds similar k-mers in the database (to exclude evolutionary influences) →
everything that scores above 12 in the substitution matrix with the original k-mer
is similar and involved in the BLAST.
- Extends seeds in both directions to find HSPs (regions that can be aligned with a
score above a certain threshold) between the query and a hit. → pairwise
sequence alignment.
➔ To only find closely related hits you make your k-mer
length longer and the neighbor threshold higher.
High scoring pair: part of the query and sequence where
the alignment score only goes up, and therefore we can say
that in this area our sequences can be related.
Local alignment: finds the optimal sub-alignment within 2 sequences. These are mostly
partial homologs (that part of the sequence has a common ancestor). → BLAST
exclusively makes local alignments.
Global alignment: aligns 2 sequences from end to end. If you know the 3 sequences are
full homologs (belong to the same organism).
BLAST input: in FASTA-format, query sequences. BLAST output: hits for the query
sequences.
BLAST forms:
- Nucleotide-nucleotide searches: blastn (finds
homologous genes in different species), megablast
(longer k-mer lengths, finds longer alignments
between very similar nucleotide sequences, best
tool for finding highly identical hits), discontiguous megablast (uses
discontiguous k-mers (excludes the 3d wobble base of an amino acid), focusses
the search on codons, best tool for finding nucleotide-nucleotide hits at larger
evolutionary distances).
- Protein-protein searches: blastp (finds homologous proteins in different species).
- Translated searches: blastx (put in nucleotides and BLAST translates your
sequences into a protein sequence, afterwards looks into protein database for
your translated sequence), tblastn (translates your protein sequence to
nucleotide sequence and look in the nucleotide database for your sequence),
tblastx (translates 2 nucleotide sequences to protein sequences and compares
the protein sequences, we do this because protein sequences are longer
conserved than nucleotide sequences).
➔ BLAST order: mask low-complexity regions – make a list of high scoring k-mers –
make a list of neighboring k-mers – make an index search with high scoring k-
mers – extend the alignment.
