GISAIDR

Programmatically interact with the GISAID EpiCoV, EpiPox, and EpiRSV databases.

Citation

If you use GISAIDR in your research please cite as:

Wytamma Wirth, & Sebastian Duchene. (2022). GISAIDR: Programmatically interact with the GISAID databases. Zenodo. https://doi.org/10.5281/zenodo.6474693

Installation

Install from github using devtools.

install.packages("devtools") # if you have not installed "devtools" package
devtools::install_github("Wytamma/GISAIDR")

Login

Get username and password from GISAID.

username = Sys.getenv("GISAIDR_USERNAME")
password = Sys.getenv("GISAIDR_PASSWORD")

Login and save your credentials (they are used for all future database queries)

credentials <- login(username = username, password = password)

Select a database

The EpiCoV database is selected by default, however, GISAIDR also works with the EpiRSV and EpiPox databases (limited testing).

credentials <- login(username = username, password = password, database="EpiRSV")
# or
credentials <- login(username = username, password = password, database="EpiPox")

Note: You need a GISAID account with access to EpiRSV and EpiPox.

Get Data

Query the database with query() using your credentials

df <- query(credentials = credentials)
head(df[0:6])

Pagination

Use nrows and start_index to page through results. GISAID limits the number of results returned with each request to 50. Internally GISAIDR runs a loop to batch queries with > 50 rows requested. See fast option below.

df <- query(credentials = credentials, nrows = 1000, start_index = 100)
nrow(df)

[1] 1000

Fast query

Use fast to load all of the accesion_ids that match the query. These accesion_ids can then be used in the download function to download up to 5000 sequences at a time.

df <- query(
  credentials = credentials, 
  location = "Oceania", 
  from_subm = "2022-07-26", 
  to_subm = "2022-07-28",
  fast = TRUE
)
head(df$accession_id)

Selecting all 484 accession_ids.
Returning 0-484 of 484 accession_ids.
[1] "EPI_ISL_14061265" "EPI_ISL_14061266" "EPI_ISL_14061267" "EPI_ISL_14061268" "EPI_ISL_14061269" "EPI_ISL_14061270"

Ordering

Use order_by to order the results or query by a column. Use order_asc to change the direction of order_by (defaults to TRUE).

df <- query(credentials = credentials, order_by = 'submission_date')
df$submission_date

[1] "2020-01-10" "2020-01-10" "2020-01-11" "2020-01-11" "2020-01-11" "2020-01-12" "2020-01-14"
[8] "2020-01-14" "2020-01-14" "2020-01-14" "2020-01-16" "2020-01-17" "2020-01-17" ...

Full text search

Use text for full text search.

accession_ids = c("EPI_ISL_17398411", "EPI_ISL_17199001", "EPI_ISL_17409201", "EPI_ISL_17243716")
df <- query(credentials = credentials, text = paste(accession_ids, collapse = "\n"))
> df$accession_id

[1] "EPI_ISL_17199001" "EPI_ISL_17243716" "EPI_ISL_17398411" "EPI_ISL_17409201"

Search by location

Use location to search for entries based on geographic location.

df <- query(credentials = credentials, location = 'Australia')
df$location

[1] "Oceania / Australia / Western Australia" "Oceania / Australia / Queensland"
[3] "Oceania / Australia / Queensland" "Oceania / Australia / Queensland"
[5] "Oceania / Australia / Western Australia" ...

A list of GISAID locations (not complete) can be found in GISAID_LOCATIONS.txt. The location search is hierarchical e.g. querying for 'Africa / ...' will return all the regions within Africa, while querying for 'Africa / Angola / ...' will only return the regions in Angola. Region can be further subdivided by specifying more levels e.g. 'North America / USA / Alabama / Butler County'. The search uses pattern matching and does not have to follow the hierarchical format above.

Search by lineage (EpiCoV)

Use lineage to search for entries based on pango lineage designations.

df <- query(credentials = credentials, lineage = 'B.1.1.7')
full_df <- download(credentials = credentials, list_of_accession_ids = df$accession_id)  # see below for download() info.
full_df$pangolin_lineage

[1] "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7"
[11] "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7" "B.1.1.7"
[21] ...

Search by Variant (EpiCoV)

Variants can be queried by name e.g. 'omicron', 'gh/490r', 'delta', 'alpha', 'beta', 'gamma', 'lambda', or 'mu'. Unfortunately GISAID doesn't return the variant designation from the query or download so variants must be confirmed with pangolin_lineage or GISAID_clade.

# VOC Omicron GRA (B.1.1.529+BA.*) first detected in Botswana/Hong Kong/South Africa
omicron_df <- query(credentials = credentials, variant = 'omicron')
omicron_full_df <- download(credentials = credentials, list_of_accession_ids = omicron_df$accession_id)
omicron_full_df$pangolin_lineage

[1] "BA.2" "BA.2" "BA.2.10.1" "BA.2" "BA.2" "BA.5" "BA.2" "BA.2" "BA.2.3"
[10] "BA.4" "BA.2" "BA.2" "BA.2" "BA.2" "BA.2" "BA.2" "BA.2" "BA.1.17"
[19] ...

Search by collection date

Use from and to to search for entries from specific dates.

df <- query(credentials = credentials, from = '2021-04-05', to = '2021-04-06')
df$collection_date

[1] "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05"
[8] "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05"
[15] ...

Search by submission date

Use from_subm and to_subm to search for entries from specific dates.

df <- query(credentials = credentials, from_subm = '2021-04-05', to_subm = '2021-04-05')
df$submission_date

[1] "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05"
[8] "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05" "2021-04-05"
[15] ...

Search by virus name

Use virus_name to search for entries using the virus name.

df <- query(credentials = credentials, virus_name="hCoV-19/Ireland/D-BHTEST/2022")
df$virus_name

[1] "hCoV-19/Ireland/D-BHTEST/2022"

Search for multiple virus names using a list.

virus_names <- list("hCoV-19/Ireland/KY-Enfer-230922004_A4/2022", "hCoV-19/Ireland/CO-Enfer-240922010_E9/2022")
df <- query(credentials = credentials, virus_name=virus_names)
df$virus_name

[1] "hCoV-19/Ireland/CO-Enfer-240922010_E9/2022" "hCoV-19/Ireland/KY-Enfer-230922004_A4/2022"

You can also match parts of the virus name e.g.

df <- query(credentials = credentials, virus_name="hCoV-19/Ireland")
df$virus_name

[1] "hCoV-19/Ireland/KY-Enfer-260922007_C6/2022" "hCoV-19/Ireland/KY-Enfer-260922007_C4/2022"
[3] "hCoV-19/Ireland/KY-Enfer-260922007_C2/2022" "hCoV-19/Ireland/KY-Enfer-260922007_C10/2022"
[5] "hCoV-19/Ireland/KY-Enfer-260922007_C1/2022" "hCoV-19/Ireland/CO-Enfer-260922007_B7/2022"...

Search by AA Substitutions and Nucleotide Mutations

Use aa_substitution and nucl_mutation to search for entries using amino acid Substitutions and nucleotide mutations.

aa_substitution_df <- query(credentials = credentials, aa_substitution = 'Spike_E484Q, Spike_H69del, -N_P13L')
nucl_mutation_df <- query(credentials = credentials, nucl_mutation = '-T23599G, -C10029T')

Exclude low coverage entries

Use low_coverage_excl to exclude low coverage entries from the results.

df <- query(credentials = credentials, low_coverage_excl = TRUE)
grep("Long stretches of NNNs", df$information)

integer(0)

Include only complete entries

GISAID considers genomes >29,000 nt as complete. Use complete to include only complete entries in the results.

df <- query(credentials = credentials, complete = TRUE)
all(df$length > 29000)

[1] TRUE

Include only high coverage entries

GISAID considers genomes with <1% Ns and <0.05% unique amino acid mutations as high coverage . Use high_coverage to include only high coverage entries in the results.

df <- query(credentials = credentials, high_coverage = TRUE)
length(grep("warn_sign", df$information)) == 0

[1] TRUE

Include only entries with complete collection date

Use collection_date_complete to include only entries with complete collection date.

df <- query(credentials = credentials, collection_date_complete = TRUE)

Load all entries

Use load_all to get all the entries that match your query without having to specify nrows.

df <- query(credentials = credentials, lineage = 'W.1', load_all = TRUE)
nrow(df)

[1] 100

Note: you may end up downloading the entire GISAID database if your query is too general.

Get total query matches

Use total to get the number of entries that match you query.

total <- query(credentials = credentials, total = TRUE)
total

[1] 10145747

Download

To download the full data set you need a list of accession IDs (which can be obtained from query results). This will also download the sequence data for each entry.

full_df_with_seq <- download(
    credentials = credentials, 
    list_of_accession_ids = list_of_accession_ids, 
)
full_df_with_seq$sequence

[1] "AGATCTGTTCTCTAAACGAACTTTAAAATCT...
[2] "AGATCTGTTCTCTAAACGAACTTTAAAATCT...
[3] "AGATCTGTTCTCTAAACGAACTTTAAAATCT...
...

You can stop GISAIDR from loading the sequence data into the memory by setting get_sequence=FALSE. Note: the sequence data will still be downloaded.

df <- query(credentials = credentials)
list_of_accession_ids <- df$accession_id
full_df <- download(credentials = credentials, list_of_accession_ids = list_of_accession_ids, get_sequence=FALSE)
colnames(full_df)

[1] "strain" "virus" "gisaid_epi_isl" "genbank_accession"
[5] "date" "region" "country" "division"
[9] "location" "region_exposure" "country_exposure" "division_exposure"
[13] "segment" "length" "host" "age"
[17] "sex" "Nextstrain_clade" "pangolin_lineage" "GISAID_clade"
[21] "originating_lab" "submitting_lab" "authors" "url"
[25] "title" "paper_url" "date_submitted" "purpose_of_sequencing"

Note: a maximum of 5000 results can be downloaded at a time.

Export to fasta file

Use the export_fasta function to write sequence data to a file in fasta format. The sequence names will be country@pango_lineage@accesion_id@date, with the date in decimal format (requires the lubridate package). The default is to only export sequences for which a decimal date could be set. To prevent this, use the argument export_dated_only = F.

export_fasta(full_df_with_seq, out_file_name = 'GISAID_sequences.fasta')

Date format (default: decimal year) and delimiter (default: @) can be set with the date_format and delimiter arguments respectively.

export_fasta(full_df_with_seq, out_file_name = 'GISAID_sequences.fasta', date_format='%Y-%m-%d', delimiter='|')

Use the columns argument to choose which columns are included in the export.

export_fasta(full_df_with_seq, out_file_name = 'GISAID_sequences.fasta', columns = c("accession_id", "country", "pangolin_lineage", "date"))

Errors

GISAIDR relies on the custom selection interface of gisaid.org. If GISAIDR is giving you errors, first check that it is not gisaid.org producing these errors. We can't do anything to fix errors with gisaid.org.

If you have an epiflu account (i.e. you were using GISAID before COVID-19) you may have issues logging in as GISAID may default you to the epiflu database.

Updating

When updating GISAIDR run detach("package:GISAIDR", unload=TRUE) first to ensure the update is applied.

Examples

Download all of the 2020 entries from Asia but outside China.

library(GISAIDR)

# country list from GISAID
Asia <- c('Asia / Afghanistan',
 'Asia / Armenia',
 'Asia / Bahrain',
 'Asia / Bangladesh',
 'Asia / Brunei',
 'Asia / Cambodia',
 # 'Asia / China', # remove China
 'Asia / Georgia',
 'Asia / Hong Kong',
 'Asia / India',
 'Asia / Indonesia / ...',
 'Asia / Iran',
 'Asia / Iraq',
 'Asia / Israel',
 'Asia / Japan',
 'Asia / Jordan / ...',
 'Asia / Kazakhstan / ...',
 'Asia / Kuwait',
 'Asia / Lebanon',
 'Asia / Malaysia',
 'Asia / Mongolia',
 'Asia / Myanmar',
 'Asia / Nepal',
 'Asia / Oman / ...',
 'Asia / Pakistan',
 'Asia / Palestine / ...',
 'Asia / Philippines',
 'Asia / Qatar / ...',
 'Asia / Saudi Arabia',
 'Asia / Singapore',
 'Asia / South Korea',
 'Asia / Sri Lanka',
 'Asia / Taiwan',
 'Asia / Thailand',
 'Asia / Timor-Leste',
 'Asia / United Arab Emirates',
 'Asia / Uzbekistan',
 'Asia / Vietnam')

credentials <- login(username = username, password = password)

asia_not_china_df <- data.frame()
for (country in Asia) {
  print(country)
  df <- query(credentials = credentials, location = country, load_all = TRUE, from = "2020-01-01", to = "2020-12-31")
  asia_not_china_df <- rbind(asia_not_china_df, df)
}
head(asia_not_china_df)

Dev guide

1. Go to the search interface on https://www.epicov.org/epi3/frontend (EpiCoV > Seach).
2. Right click on the feature you want to add (e.g. the complete checkbox) and inspect the source code.
3. Find the ceid for this element (<div id="ce_qxos9a_bi">) e.g. ce_qxos9a_bi
4. Find the the value of the checkbox element (<input class="sys-event-hook" name="ce_qxos9a_bi_name" style="vertical-align: middle;" type="checkbox" value="complete">) e.g. complete.
5. Search for the id in the page source. In one of the <script> header tags you'll find a createFI() function e.g. this.getForm().createFI('ce_qxos9a_bi','CheckboxWidget','quality'. The ceid is dynamic so you need to use the widget name i.e. quality to find the ceid dynamically.
6. In the GISAIDR login.R file add code to extract the ceid using the extract_search_ceid() function and the widget e.g.

# Complete 
quality_ceid <- extract_search_ceid("quality'", customSearch_page_text)

1. Add the extracted ceid to the list of credentials e.g. complete_ceid = complete_ceid
2. Add the new argument and default value to all query() function in query.R and internal_query.R e.g. complete = FALSE.
3. Add the new argument and default value to the (load_all && j$totalRecords > nrows) load_all recursion loop so all paginations will continue using the argument.
4. Create and append a search queue to the main queue if the complete argument is used. Create the command using the create_search_queue() function. Use the complete_ceid for the ceid and the checkbox value (identified in step 4) for the cvalue e.g.

if (complete) {
  queue <-
    append(
      queue,
      create_search_queue(credentials,
                          ceid=credentials$complete_ceid,
                          cvalue=list('complete'),
                          cmd='FilterChange')
    )
}

1. Add the new argument to the recursive load all query() call e.g. complete = complete
2. Add test for new feature in test-query.R e.g.

test_that("complete works", {
  df <- query(credentials = credentials, complete = TRUE)
  expect_true(all(df$length > 29000))
})

1. Run all tests. If tests pass create a PR to merge the new feature.

See this PR for the full details.