Cheminformatics search and AMOS

Cheminformatics searches have different input contracts and result shapes. Start with one search type and inspect its result before you combine searches.

Search chemical space

library(ComptoxR)
library(dplyr)

exact <- chemi_search("DTXSID7020182", search_type = "exact")
similar <- chemi_search(
  "DTXSID7020182",
  search_type = "similar",
  min_similarity = 0.9
)
substructure <- chemi_search("c1ccccc1", search_type = "substructure")
mass <- chemi_search(
  search_type = "mass",
  mass_type = "mono",
  min_mass = 77,
  max_mass = 79
)
features <- chemi_search(
  search_type = "features",
  element_include = c("C", "H"),
  exclude_all_others = TRUE
)

Exact, similarity, and substructure searches require a structure or chemical query. Mass search requires a mass type and range. Feature search uses element and structure flags.

Hazard search uses service categories instead of a structure query.

hazard <- chemi_search(
  search_type = "hazard",
  hazard_name = "cancer",
  min_toxicity = "H",
  min_authority = "auth"
)

Predict and cluster

prediction <- chemi_predictor_models_predict(
  model_id = 1065,
  identifier = "DTXSID7020182"
)

clusters <- chemi_resolver_getsimilaritymap(
  c("DTXSID7020182", "DTXSID0020319", "DTXSID9020243"),
  idType = "DTXSID",
  format = "cluster"
)
plot(clusters$hc)

Model IDs belong to models deployed by the service. A known ID can be removed, replaced, or unavailable on the selected server. Record the model ID and service environment with each prediction. Treat availability as service-controlled, not as a package guarantee.

Inspect AMOS analytical methods

AMOS method discovery returns analytical method records. It is not a chemical lookup.

methods <- chemi_amos_method_list()
glimpse(methods)
head(methods)

Do not use chemi_amos_method_with_spectra() for a compound lookup. It requires an AMOS method or spectrum record ID.

Move through ClassyFire categories

First get the four ClassyFire levels for benzene.

classification <-
  chemi_amos_get_classification_for_dtxsid("DTXSID7020182") |>
  bind_rows()

classification |>
  select(any_of(c("kingdom", "superklass", "klass", "subklass")))

Then enumerate valid child categories. Use klass; class is not a supported request parameter.

classes <- chemi_amos_next_level_classification(
  kingdom = classification$kingdom[[1]],
  superklass = classification$superklass[[1]]
)

subclasses <- chemi_amos_next_level_classification(
  kingdom = classification$kingdom[[1]],
  superklass = classification$superklass[[1]],
  klass = classification$klass[[1]]
)

Move from one ClassyFire category back to matching substances.

class_members <- chemi_amos_substances_for_classification(
  kingdom = classification$kingdom[[1]],
  superklass = classification$superklass[[1]],
  klass = classification$klass[[1]],
  subklass = classification$subklass[[1]]
)

Request ClassyFire fields for several DTXSIDs with the AMOS batch workflow.

amos_batch <- chemi_amos_batch(
  dtxsids = c("DTXSID7020182", "DTXSID0020319", "DTXSID9020243"),
  include_classyfire = TRUE
)
glimpse(amos_batch)

ClassyFire describes chemical structure classes. ChemFuncT describes functional uses. Keep the two classification systems separate. Advanced users can query ChemFuncT with chemi_amos_functional_uses_for_dtxsid("DTXSID7020182").