Orthology Inference using orthologr
2026-08-10
Source:vignettes/orthology_inference.Rmd
orthology_inference.RmdOrthology Inference is the process of detecting orthologous genes between a
query organism and a set of subject organisms. Motivated by the
discussion on which orthology inference method, paradigm or program is
the most accurate or the fastest one,
the orthologr package provides functions to perform genome
wide orthology inference following two different paradigms:
orthology inference based on 1-1 orthology relationship using
DIAMOND2 reciprocal best hit(default, recommended) orBLAST reciprocal best hitorthology inference based on 1-many homology relationship using orthologous groups (OrthoFinder2)
To perform orthology inference you can start with the
orthologs() function provided by orthologr.
The orthologs() function takes nucleotide or protein
sequences stored in fasta files for a set of organisms as input and
performs orthology inference to detect orthologous genes or orthologous
groups within the given organisms based on selected orthology inference
programs.
The following interfaces are implemented in the
orthologs() function:
DIAMOND2 based methods (recommended):
DIAMOND2 reciprocal best hit (DIAMOND_RBH): see
diamond_rec(). This is the default (ortho_detection = "DIAMOND_RBH"). DIAMOND2 is up to 10,000× faster than BLAST in default mode and retains full BLAST-level sensitivity inultra-sensitivemode.DIAMOND2 best hit (DIAMOND_BH): see
diamond_best().
BLAST based methods:
BLASTp best hit (BH): see
blast_best().BLASTp reciprocal best hit (RBH): see
blast_rec().
Orthogroup based methods:
-
Orthofinder2:
see
orthofinder2().
Examples: DIAMOND2 based methods (default)
Using a simple example stored in the package environment of
orthologr you can get an impression on how to use the
orthologs() function.
Note: it is assumed that when using
orthologs() all corresponding programs you want to use are
already installed on your machine and are executable via either the
default execution PATH or you specifically define the location of the
executable file via the path argument that can be passed to
orthologs(). For DIAMOND2-based methods (the default), make
sure DIAMOND2 is
installed. See the Installation
Vignette for details.
In the following examples, we will use toy fasta files stored within
the orthologr package:
system.file('seqs/ortho_thal_aa.fasta', package = 'orthologr').
# perform orthology inference using DIAMOND2 reciprocal best hit (default)
# and fasta sequence files storing protein sequences
orthologr::orthologs(query_file = system.file('seqs/ortho_thal_aa.fasta', package = 'orthologr'),
subject_files = system.file('seqs/ortho_lyra_aa.fasta', package = 'orthologr'),
seq_type = "protein",
ortho_detection = "DIAMOND_RBH",
comp_cores = 1,
clean_folders = FALSE)Running diamond version 2.2.1 ...
sensitivity mode: fast
creating a diamond database
Starting DIAMOND2 search ...
DIAMOND2 search completed in 0.02 sec.
Running diamond version 2.2.1 ...
sensitivity mode: fast
creating a diamond database
Starting DIAMOND2 search ...
DIAMOND2 search completed in 0.02 sec.
# A tibble: 20 × 21
# Groups: query_id [20]
query_id subject_id perc_identity num_ident_matches alig_length
<chr> <chr> <dbl> <int> <int>
1 AT1G01010.1 333554|PACid:160… 73.2 347 474
2 AT1G01020.1 470181|PACid:160… 91.1 224 246
3 AT1G01030.1 470180|PACid:160… 93.3 335 359
4 AT1G01040.1 333551|PACid:160… 93.4 1840 1969
5 AT1G01050.1 909874|PACid:160… 100 213 213
6 AT1G01060.3 470177|PACid:160… 87.5 567 648
7 AT1G01070.1 918864|PACid:160… 92.6 339 366
8 AT1G01080.1 909871|PACid:160… 89.3 268 300
9 AT1G01090.1 470171|PACid:160… 96.8 420 434
10 AT1G01110.2 333544|PACid:160… 87.7 463 528
11 AT1G01120.1 918858|PACid:160… 99.2 525 529
12 AT1G01140.3 470161|PACid:160… 98.5 446 453
13 AT1G01150.1 918855|PACid:160… 72.6 207 285
14 AT1G01160.1 918854|PACid:160… 78.8 141 179
15 AT1G01170.2 311317|PACid:160… 85.6 83 97
16 AT1G01180.1 909860|PACid:160… 92.6 287 310
17 AT1G01190.1 311315|PACid:160… 94.2 502 533
18 AT1G01200.1 470156|PACid:160… 95.8 228 238
19 AT1G01210.1 311313|PACid:160… 95.3 102 107
20 AT1G01220.1 470155|PACid:160… 96.5 1019 1056
# ℹ 16 more variables: mismatches <int>, gap_openings <int>, n_gaps <int>,
# pos_match <int>, ppos <dbl>, q_start <int>, q_end <int>, q_len <int>,
# qcov <dbl>, qcovhsp <dbl>, s_start <int>, s_end <int>, s_len <int>,
# evalue <dbl>, bit_score <dbl>, score_raw <dbl>This small example returns 20 orthologous genes between
Arabidopsis thaliana and Arabidopsis lyrata. As you
can see, the query_file and subject_files
arguments take the proteomes of Arabidopsis thaliana
(query_file) and Arabidopsis lyrata
(subject_files) stored in fasta files. The
seq_type argument specifies that you will pass protein
sequences (proteomes) to the orthologs() function. In case
you only have either genomes (DNA sequences) or CDS files, you can
modify the seq_type argument to
seq_type = "dna" (when working with only genome data) or
seq_type = "cds" (when working with CDS files). Internally
the orthologs() function will perform a CDS prediction
using predict_cds() and will furthermore translate the
predicted CDS sequences into protein sequences. Analogously when
seq_type = "cds" is specified, internally the
orthologs() function will translate all CDS sequences into
protein sequences to run orthology inference based on protein
sequences.
The advantage of this type of output is that in addition to having the orthology relationship between genes from two different genomes, users also retrieve the DIAMOND2 alignment results of the respective orthology relationship (in the same format as BLAST) which allows them to perform subsequent filtering for either more conservative or more liberal orthology relationships.
Note: future versions of orthologr will
allow to perform orthology inference using DNA sequences. Nevertheless,
since most orthology inference methods or paradigms rely on protein
sequences, the first version of orthologr will follow this
paradigm.
In case you have to specify the path to DIAMOND2 you can use the
path argument as follows:
# using an external execution path for DIAMOND2
orthologr::orthologs(query_file = system.file('seqs/ortho_thal_aa.fasta', package = 'orthologr'),
subject_files = system.file('seqs/ortho_lyra_aa.fasta', package = 'orthologr'),
seq_type = "protein",
ortho_detection = "DIAMOND_RBH",
path = "here/path/to/diamond",
clean_folders = FALSE,
comp_cores = 1)When you are working on a multicore machine, you can also specify the
comp_cores argument that will allow you to run all analyses
in parallel (to speed up computations).
# running DIAMOND2 orthology inference in parallel using 2 cores
orthologr::orthologs(query_file = system.file('seqs/ortho_thal_aa.fasta', package = 'orthologr'),
subject_files = system.file('seqs/ortho_lyra_aa.fasta', package = 'orthologr'),
seq_type = "protein",
ortho_detection = "DIAMOND_RBH",
clean_folders = FALSE,
comp_cores = 2)# A tibble: 20 x 21
# Groups: query_id [20]
query_id subject_id perc_identity alig_length
<chr> <chr> <dbl> <dbl>
1 AT1G010... 333554|PA... 74.0 469
2 AT1G010... 470181|PA... 91.1 246
3 AT1G010... 470180|PA... 95.5 359
4 AT1G010... 333551|PA... 92.0 1970
5 AT1G010... 909874|PA... 100 213
6 AT1G010... 470177|PA... 89.5 648
7 AT1G010... 918864|PA... 95.1 366
8 AT1G010... 909871|PA... 90.3 300
9 AT1G010... 470171|PA... 96.8 434
10 AT1G011... 333544|PA... 93.6 528
11 AT1G011... 918858|PA... 99.2 529
12 AT1G011... 470161|PA... 98.4 453
13 AT1G011... 918855|PA... 72.6 285
14 AT1G011... 918854|PA... 84.9 179
15 AT1G011... 311317|PA... 85.6 97
16 AT1G011... 909860|PA... 92.6 310
17 AT1G011... 311315|PA... 94.2 533
18 AT1G012... 470156|PA... 95.8 238
19 AT1G012... 311313|PA... 95.3 107
20 AT1G012... 470155|PA... 96.7 1056
# ... with 17 more variables: mismatches <dbl>,
# gap_openings <dbl>, n_gaps <dbl>, pos_match <dbl>, ppos <dbl>,
# q_start <dbl>, q_end <dbl>, q_len <dbl>, qcov <dbl>, qcovhsp <dbl>,
# s_start <dbl>, s_end <dbl>, s_len <dbl>, evalue <dbl>,
# bit_score <dbl>, score_raw <dbl>In this case 2 cores are being used to perform parallel processing,
the clean_folders argument specifies that all files
returned by the corresponding orthology inference method are removed
after analyses.
Program specific use of the orthologs() function
In this section small examples will illustrate the use of the
orthologs() function for each orthology inference
program.
DIAMOND2 best hit (recommended)
The DIAMOND2 best hit method is a uni-directional DIAMOND2 best hit
search of a query organism A against a
subject organism B based on the e-value. It
uses the same hit-filtering logic as blast_best() but is
substantially faster.
Orthology Inference using DIAMOND2 best hit can be performed by
specifying the argument ortho_detection = "DIAMOND_BH" and
one computing core comp_core = 1:
# perform orthology inference using DIAMOND2 best hit
# and fasta sequence files storing protein sequences
orthologr::orthologs(query_file = system.file('seqs/ortho_thal_aa.fasta', package = 'orthologr'),
subject_files = system.file('seqs/ortho_lyra_aa.fasta', package = 'orthologr'),
seq_type = "protein",
ortho_detection = "DIAMOND_BH",
clean_folders = TRUE,
comp_cores = 1)# A tibble: 20 × 21
# Groups: query_id [20]
query_id subject_id perc_identity num_ident_matches alig_length
<chr> <chr> <dbl> <int> <int>
1 AT1G01010.1 333554|PAC… 73.2 347 474
2 AT1G01020.1 470181|PAC… 91.1 224 246
3 AT1G01030.1 470180|PAC… 93.3 335 359
4 AT1G01040.1 333551|PAC… 93.4 1840 1969
5 AT1G01050.1 909874|PAC… 100 213 213
6 AT1G01060.3 470177|PAC… 87.5 567 648
7 AT1G01070.1 918864|PAC… 92.6 339 366
8 AT1G01080.1 909871|PAC… 89.3 268 300
9 AT1G01090.1 470171|PAC… 96.8 420 434
10 AT1G01110.2 333544|PAC… 87.7 463 528
11 AT1G01120.1 918858|PAC… 99.2 525 529
12 AT1G01140.3 470161|PAC… 98.5 446 453
13 AT1G01150.1 918855|PAC… 72.6 207 285
14 AT1G01160.1 918854|PAC… 78.8 141 179
15 AT1G01170.2 311317|PAC… 85.6 83 97
16 AT1G01180.1 909860|PAC… 92.6 287 310
17 AT1G01190.1 311315|PAC… 94.2 502 533
18 AT1G01200.1 470156|PAC… 95.8 228 238
19 AT1G01210.1 311313|PAC… 95.3 102 107
20 AT1G01220.1 470155|PAC… 96.5 1019 1056
# ℹ 16 more variables: mismatches <int>, gap_openings <int>,
# n_gaps <int>, pos_match <int>, ppos <dbl>, q_start <int>,
# q_end <int>, q_len <int>, qcov <dbl>, qcovhsp <dbl>,
# s_start <int>, s_end <int>, s_len <int>, evalue <dbl>,
# bit_score <dbl>, score_raw <dbl>The resulting table stores the orthologous gene pairs and the
corresponding alignment statistics. DIAMOND2 returns additional columns
compared to BLAST (e.g. n_gaps, pos_match,
ppos, q_len, qcov,
s_len, score_raw).
DIAMOND2 reciprocal best hit (default, recommended)
The DIAMOND2 reciprocal best hit is a bi-directional DIAMOND2 best
hit search of a query organism A against a
subject organism B based on the e-value. This
is the default method in orthologs().
The algorithm runs as follows:
bh_A <- diamond_best(A, B)bh_B <- diamond_best(B, A)join(bh_A, bh_B)by tuple(query_id, subject_id)
Only when the tuple (query_id, subject_id) is returned
as best hit in both DIAMOND2 directions is it retained as an orthologous
gene pair.
library(orthologr)
# perform orthology inference using DIAMOND2 reciprocal best hit (default)
# and fasta sequence files storing protein sequences
orthologs(query_file = system.file('seqs/ortho_thal_aa.fasta', package = 'orthologr'),
subject_files = system.file('seqs/ortho_lyra_aa.fasta', package = 'orthologr'),
seq_type = "protein",
ortho_detection = "DIAMOND_RBH",
clean_folders = FALSE,
comp_cores = 1)# A tibble: 20 × 21
# Groups: query_id [20]
query_id subject_id perc_identity num_ident_matches alig_length
<chr> <chr> <dbl> <int> <int>
1 AT1G01010.1 333554|PAC… 73.2 347 474
2 AT1G01020.1 470181|PAC… 91.1 224 246
3 AT1G01030.1 470180|PAC… 93.3 335 359
4 AT1G01040.1 333551|PAC… 93.4 1840 1969
5 AT1G01050.1 909874|PAC… 100 213 213
6 AT1G01060.3 470177|PAC… 87.5 567 648
7 AT1G01070.1 918864|PAC… 92.6 339 366
8 AT1G01080.1 909871|PAC… 89.3 268 300
9 AT1G01090.1 470171|PAC… 96.8 420 434
10 AT1G01110.2 333544|PAC… 87.7 463 528
11 AT1G01120.1 918858|PAC… 99.2 525 529
12 AT1G01140.3 470161|PAC… 98.5 446 453
13 AT1G01150.1 918855|PAC… 72.6 207 285
14 AT1G01160.1 918854|PAC… 78.8 141 179
15 AT1G01170.2 311317|PAC… 85.6 83 97
16 AT1G01180.1 909860|PAC… 92.6 287 310
17 AT1G01190.1 311315|PAC… 94.2 502 533
18 AT1G01200.1 470156|PAC… 95.8 228 238
19 AT1G01210.1 311313|PAC… 95.3 102 107
20 AT1G01220.1 470155|PAC… 96.5 1019 1056
# ℹ 16 more variables: mismatches <int>, gap_openings <int>,
# n_gaps <int>, pos_match <int>, ppos <dbl>, q_start <int>,
# q_end <int>, q_len <int>, qcov <dbl>, qcovhsp <dbl>,
# s_start <int>, s_end <int>, s_len <int>, evalue <dbl>,
# bit_score <dbl>, score_raw <dbl>You can also adjust the DIAMOND2 sensitivity level using the
sensitivity_mode argument. The default is
"fast", but for more distant species comparisons you may
want "sensitive" or "ultra-sensitive":
library(orthologr)
# DIAMOND2 RBH with ultra-sensitive mode (equivalent to BLAST sensitivity)
orthologs(query_file = system.file('seqs/ortho_thal_aa.fasta', package = 'orthologr'),
subject_files = system.file('seqs/ortho_lyra_aa.fasta', package = 'orthologr'),
seq_type = "protein",
ortho_detection = "DIAMOND_RBH",
sensitivity_mode = "ultra-sensitive",
clean_folders = FALSE,
comp_cores = 1)The full set of sensitivity modes available in DIAMOND2 is:
sensitivity_mode |
Speed | Use case |
|---|---|---|
"faster" |
Fastest | Hits >70% identity |
"fast" |
Fast (default) | Hits >70% identity |
"mid-sensitive" |
Moderate | Between fast and sensitive |
"sensitive" |
Sensitive | Full sensitivity for hits >40% identity |
"more-sensitive" |
More sensitive | — |
"very-sensitive" |
Very sensitive | — |
"ultra-sensitive" |
BLAST-equivalent | Most sensitive; slowest |
To retrieve the full DIAMOND2 hit table for subsequent analyses, set
clean_folders = FALSE. The corresponding hit table can then
be found in file.path(tempdir(), "_blast_db").
library(orthologr)
library(dplyr)
# perform orthology inference using DIAMOND2 reciprocal best hit
RBH <- orthologs(query_file = system.file('seqs/ortho_thal_aa.fasta', package = 'orthologr'),
subject_files = system.file('seqs/ortho_lyra_aa.fasta', package = 'orthologr'),
seq_type = "protein",
ortho_detection = "DIAMOND_RBH",
clean_folders = FALSE,
comp_cores = 1)
glimpse(RBH)Rows: 20
Columns: 21
Groups: query_id [20]
$ query_id <chr> "AT1G01010.1", "AT1G01020.1", "AT1G01030.1…
$ subject_id <chr> "333554|PACid:16033839", "470181|PACid:160…
$ perc_identity <dbl> 73.2, 91.1, 93.3, 93.4, 100.0, 87.5, 92.6,…
$ num_ident_matches <int> 347, 224, 335, 1840, 213, 567, 339, 268, 4…
$ alig_length <int> 474, 246, 359, 1969, 213, 648, 366, 300, 4…
$ mismatches <int> 75, 22, 20, 58, 0, 71, 23, 25, 8, 65, 4, 6…
$ gap_openings <int> 8, 0, 2, 7, 0, 5, 2, 2, 3, 0, 0, 1, 3, 2, …
$ n_gaps <int> 52, 0, 4, 71, 0, 10, 4, 7, 6, 0, 0, 1, 10,…
$ pos_match <int> 369, 231, 338, 1870, 213, 586, 342, 275, 4…
$ ppos <dbl> 77.8, 93.9, 94.2, 95.0, 100.0, 90.4, 93.4,…
$ q_start <int> 1, 1, 1, 6, 1, 1, 1, 1, 1, 1, 1, 1, 5, 4, …
$ q_end <int> 430, 246, 359, 1910, 213, 646, 366, 294, 4…
$ q_len <int> 430, 246, 359, 1910, 213, 646, 366, 294, 4…
$ qcov <dbl> 100.0, 100.0, 100.0, 99.7, 100.0, 100.0, 1…
$ qcovhsp <dbl> 100.0, 100.0, 100.0, 99.7, 100.0, 100.0, 1…
$ s_start <int> 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 16, 2,…
$ s_end <int> 466, 246, 355, 1963, 213, 640, 362, 299, 4…
$ s_len <int> 466, 246, 355, 1963, 213, 640, 362, 299, 4…
$ evalue <dbl> 1.37e-211, 2.27e-156, 4.01e-210, 0.00e+00,…
$ bit_score <dbl> 582, 426, 571, 3544, 427, 1028, 614, 494, …
$ score_raw <dbl> 1500, 1094, 1471, 9190, 1098, 2658, 1583, …BLASTp best hit
The BLASTp best hit method is a uni-directional BLAST best hit search
of a query organism A against a
subject organism B based on the e-value.
Orthology Inference using BLASTp best hit can be performed by
specifying the argument ortho_detection = "BH" and one
computing core comp_core = 1:
# perform orthology inference using BLAST best hit
# and fasta sequence files storing protein sequences
orthologr::orthologs(query_file = system.file('seqs/ortho_thal_aa.fasta', package = 'orthologr'),
subject_files = system.file('seqs/ortho_lyra_aa.fasta', package = 'orthologr'),
seq_type = "protein",
ortho_detection = "BH",
clean_folders = TRUE,
comp_cores = 1)# A tibble: 20 × 21
# Groups: query_id [20]
query_id subject_id perc_identity num_ident_matches alig_length
<chr> <chr> <dbl> <int> <int>
1 AT1G01010.1 333554|PAC… 74.0 347 469
2 AT1G01020.1 470181|PAC… 91.1 224 246
3 AT1G01030.1 470180|PAC… 95.5 343 359
4 AT1G01040.1 333551|PAC… 92.0 1812 1970
5 AT1G01050.1 909874|PAC… 100 213 213
6 AT1G01060.3 470177|PAC… 89.5 580 648
7 AT1G01070.1 918864|PAC… 95.1 348 366
8 AT1G01080.1 909871|PAC… 90.3 271 300
9 AT1G01090.1 470171|PAC… 96.8 420 434
10 AT1G01110.2 333544|PAC… 93.6 494 528
11 AT1G01120.1 918858|PAC… 99.2 525 529
12 AT1G01140.3 470161|PAC… 98.5 446 453
13 AT1G01150.1 918855|PAC… 72.6 207 285
14 AT1G01160.1 918854|PAC… 84.9 152 179
15 AT1G01170.2 311317|PAC… 85.6 83 97
16 AT1G01180.1 909860|PAC… 92.6 287 310
17 AT1G01190.1 311315|PAC… 94.2 502 533
18 AT1G01200.1 470156|PAC… 95.8 228 238
19 AT1G01210.1 311313|PAC… 95.3 102 107
20 AT1G01220.1 470155|PAC… 96.7 1021 1056
# ℹ 16 more variables: mismatches <int>, gap_openings <int>,
# n_gaps <int>, pos_match <int>, ppos <dbl>, q_start <int>,
# q_end <int>, q_len <int>, qcov <dbl>, qcovhsp <dbl>,
# s_start <int>, s_end <int>, s_len <int>, evalue <dbl>,
# bit_score <dbl>, score_raw <dbl>The resulting table stores the orthologous gene pairs and the corresponding BLAST alignment statistics.
BLASTp best reciprocal hit
The BLAST best reciprocal hit is a bi-directional BLAST best hit
search of a query organism A against a
subject organism B based on the e-value.
The Algorithm for BLAST best reciprocal hit runs as follows:
bh_A <- best_hit(A,B)bh_B <- best_hit(B,A)join(bh_A,bh_B)by tupel(query_id, subject_id)
In other words, only in case the tuple
(query_id, subject_id) is returned as best hit based on the
e-value in both BLAST directions, the corresponding tupel
(query_id, subject_id) is retained as orthologous gene
pair.
library(orthologr)
# perform orthology inference using BLAST reciprocal best hit
# and fasta sequence files storing protein sequences
orthologs(query_file = system.file('seqs/ortho_thal_aa.fasta', package = 'orthologr'),
subject_files = system.file('seqs/ortho_lyra_aa.fasta', package = 'orthologr'),
seq_type = "protein",
ortho_detection = "RBH",
clean_folders = FALSE,
comp_cores = 1)# A tibble: 20 × 21
# Groups: query_id [20]
query_id subject_id perc_identity num_ident_matches alig_length
<chr> <chr> <dbl> <int> <int>
1 AT1G01010.1 333554|PAC… 74.0 347 469
2 AT1G01020.1 470181|PAC… 91.1 224 246
3 AT1G01030.1 470180|PAC… 95.5 343 359
4 AT1G01040.1 333551|PAC… 92.0 1812 1970
5 AT1G01050.1 909874|PAC… 100 213 213
6 AT1G01060.3 470177|PAC… 89.5 580 648
7 AT1G01070.1 918864|PAC… 95.1 348 366
8 AT1G01080.1 909871|PAC… 90.3 271 300
9 AT1G01090.1 470171|PAC… 96.8 420 434
10 AT1G01110.2 333544|PAC… 93.6 494 528
11 AT1G01120.1 918858|PAC… 99.2 525 529
12 AT1G01140.3 470161|PAC… 98.5 446 453
13 AT1G01150.1 918855|PAC… 72.6 207 285
14 AT1G01160.1 918854|PAC… 84.9 152 179
15 AT1G01170.2 311317|PAC… 85.6 83 97
16 AT1G01180.1 909860|PAC… 92.6 287 310
17 AT1G01190.1 311315|PAC… 94.2 502 533
18 AT1G01200.1 470156|PAC… 95.8 228 238
19 AT1G01210.1 311313|PAC… 95.3 102 107
20 AT1G01220.1 470155|PAC… 96.7 1021 1056
# ℹ 16 more variables: mismatches <int>, gap_openings <int>,
# n_gaps <int>, pos_match <int>, ppos <dbl>, q_start <int>,
# q_end <int>, q_len <int>, qcov <dbl>, qcovhsp <dbl>,
# s_start <int>, s_end <int>, s_len <int>, evalue <dbl>,
# bit_score <dbl>, score_raw <dbl>
In case you would like to store the corresponding
hit tables returned by BLAST for subsequent analyses, you
can specify the clean_folders = FALSE argument. The
corresponding BLAST hit table can then be found in
file.path(tempdir(),"_blast_db").
A detailed overview of further analyses that can be done with the corresponding BLAST output can be found in the BLAST vignette.