Add tests and update docs

Make it clear in the docs and the help screen that when using mmseqs,
users can provide a premade target DB and it will work fine.  Also
include new tests to make sure this works okay.

README.md

@@ -105,6 +105,25 @@ A couple of things to note here:
 
 For more info on command line usage, see the help screen by running `cerise --help`.
 
+### Premade target databases
+
+If you use `mmseqs`, you can use a premade target database.  Simply pass in the path of the target DB and it will work fine.  Here is an example where I make a search DB first, then run Cerise as before.
+
+```
+$ mmseqs createdb clustered_targets.fasta clustered_targets.db
+$ cerise \
+    clustered_queries.fasta \
+    clustered_targets.db \
+    --query-clusters=query_clusters.tsv \
+    --target-clusters=target_clusters.tsv \
+    --all-queries=queries.fasta \
+    --all-targets=targets.fasta \
+    --search-config='--threads 4 -s 7 --num-iterations 3' \
+    --search-program=mmseqs
+```
+
+*Using premade DBs in this way is not yet supported when using `blast` or `diamond`.*
+
 ### More examples
 
 There are a ton of examples on how to run (and how to break) Cerise in the [test](https://github.com/mooreryan/cerise/tree/main/cerise/test) directory of this repository.  In this directory, you will see directories ending in `*.t`.  Each of these specifies a self contained set of tests, including all the files needed to run tests in that directory.  You will also find a `run.t` in each of the subdirectories.  These files are where you will find the examples exercising the behavior of the `cerise` command line app.  See [here](https://bitheap.org/cram/) for info on how to read these files.

cerise/lib/cli.ml

@@ -52,7 +52,12 @@ let queries_term =
   Arg.(required & pos 0 (some non_dir_file) None & info [] ~docv:"QUERIES" ~doc)
 
 let targets_term =
-  let doc = "Path to target sequences" in
+  let doc =
+    "Path to target sequences/db.  If you use the 'mmseqs' option, this can \
+     either be the path to sequenecs in a FASTA file, or a MMseqs2 sequence \
+     database.  If you use an already existing database, then it will skip the \
+     database construction step."
+  in
   Arg.(required & pos 1 (some non_dir_file) None & info [] ~docv:"TARGETS" ~doc)
 
 let outdir_term =

cerise/lib/clusters.ml

@@ -44,7 +44,7 @@ let maybe_keep_cluster_members ~centroid ~clusters ~keep =
 (* After the first homology search, run this to figure out which seqs to use in
    the next round. *)
 let get_new_search_input_seq_ids ~query_clusters ~target_clusters btab_fname =
-  let init =
+  let clusters =
     match (query_clusters, target_clusters) with
     | None, None ->
         failwith "you should provide either query or target clusters"
@@ -53,13 +53,18 @@ let get_new_search_input_seq_ids ~query_clusters ~target_clusters btab_fname =
     | Some _, Some _ ->
         (Some (Set.empty (module String)), Some (Set.empty (module String)))
   in
-  In_channel.with_file btab_fname ~f:(fun ic ->
-      In_channel.fold_lines ic ~init
-        ~f:(fun (keep_queries, keep_targets) line ->
-          match String.split ~on:'\t' line with
-          | query :: target :: _rest ->
-              ( maybe_keep_cluster_members ~centroid:query
-                  ~clusters:query_clusters ~keep:keep_queries,
-                maybe_keep_cluster_members ~centroid:target
-                  ~clusters:target_clusters ~keep:keep_targets )
-          | _ -> failwith "bad btab file"))
+  let count, clusters =
+    In_channel.with_file btab_fname ~f:(fun ic ->
+        In_channel.fold_lines ic ~init:(0, clusters)
+          ~f:(fun (i, (keep_queries, keep_targets)) line ->
+            match String.split ~on:'\t' line with
+            | query :: target :: _rest ->
+                ( i + 1,
+                  ( maybe_keep_cluster_members ~centroid:query
+                      ~clusters:query_clusters ~keep:keep_queries,
+                    maybe_keep_cluster_members ~centroid:target
+                      ~clusters:target_clusters ~keep:keep_targets ) )
+            | _ -> failwith "bad btab file"))
+  in
+  if count > 0 then clusters
+  else Utils.abort "ERROR: there were no hits of your queries to your targets"

cerise/lib/utils.ml

@@ -1,5 +1,9 @@
 open! Core
 
+let abort ?(exit_code = 1) msg =
+  let () = eprintf "%s\n" msg in
+  Caml.exit exit_code
+
 (* See
    https://github.com/ocaml/dune/commit/154272b779fe8943a9ce1b4afabb30150ab94ba6 *)
 

cerise/test/mmseqs_search.t/run.t

@@ -97,3 +97,94 @@ Bad target clusters.
   $ grep -A1 Failure cerise_oe | sed -E 's/^ +//'
   (Failure "bad line in clusters file 'apple\tpie\tgood'")
 
+
+Trying to do clustered targets but missing --all-targets.
+
+  $ if [ -d cerise_out ]; then rm -r cerise_out; fi
+  $ cerise -vv clustered_queries.fasta clustered_targets.fasta --target-clusters target_clusters.tsv --search-config='-s=1 --threads=4' 2> err
+  [2]
+  $ grep -A1 Failure err | sed -E 's/^ +//'
+  (Failure
+  "--target-clusters and --all-targets must both be present, or neither should be present")
+
+Trying to do clustered targets but missing --target-clusters.
+
+  $ if [ -d cerise_out ]; then rm -r cerise_out; fi
+  $ cerise -vv clustered_queries.fasta clustered_targets.fasta --all-targets targets.fasta --search-config='-s=1 --threads=4' 2> err
+  [2]
+  $ grep -A1 Failure err | sed -E 's/^ +//'
+  (Failure "you need to have at least one of query or target clusters")
+
+
+Trying to do clustered queries but missing --all-queries.
+
+  $ if [ -d cerise_out ]; then rm -r cerise_out; fi
+  $ cerise -vv clustered_queries.fasta clustered_targets.fasta --query-clusters query_clusters.tsv --search-config='-s=1 --threads=4' 2> err
+  [2]
+  $ grep -A1 Failure err | sed -E 's/^ +//'
+  (Failure
+  "--query-clusters and --all-queries must both be present, or neither should be present")
+
+Trying to do clustered queries but missing --query-clusters.
+
+  $ if [ -d cerise_out ]; then rm -r cerise_out; fi
+  $ cerise -vv clustered_queries.fasta clustered_targets.fasta --all-queries queries.fasta --search-config='-s=1 --threads=4' 2> err
+  [2]
+  $ grep -A1 Failure err | sed -E 's/^ +//'
+  (Failure "you need to have at least one of query or target clusters")
+
+
+Premade target DB; both queries and targets clustered.
+
+  $ if [ -d cerise_out ]; then rm -r cerise_out; fi
+  $ mmseqs createdb clustered_targets.fasta clustered_targets.db > /dev/null 2>&1
+  $ cerise -vv clustered_queries.fasta clustered_targets.db --query-clusters query_clusters.tsv --target-clusters target_clusters.tsv --all-queries queries.fasta --all-targets targets.fasta --search-config='-s=1 --threads=4' > cerise_oe 2>&1
+  $ ls cerise_out
+  cerise.first_search.tsv
+  cerise.new_queries.fasta
+  cerise.new_targets.fasta
+  cerise.second_search.tsv
+  command_logs.txt
+  $ grep '^>' cerise_out/cerise.new_queries.fasta | cut -f1 -d' ' | sort | diff - expected_new_queries__both.txt
+  $ grep '^>' cerise_out/cerise.new_targets.fasta | cut -f1 -d' ' | sort | diff - expected_new_targets__both.txt
+  $ sort -k1,2 cerise_out/cerise.first_search.tsv | cut -f1,2 | diff - expected_first_search__both.tsv
+  $ sort -k1,2 cerise_out/cerise.second_search.tsv | cut -f1,2 | diff - expected_second_search__both.tsv
+
+Premade target DB; just queries clustered.
+
+  $ if [ -d cerise_out ]; then rm -r cerise_out; fi
+  $ mmseqs createdb clustered_targets.fasta clustered_targets.db > /dev/null 2>&1
+  $ cerise clustered_queries.fasta clustered_targets.db --query-clusters query_clusters.tsv --all-queries queries.fasta --search-config='-s=1 --threads=4' > cerise_oe 2>&1
+  $ ls cerise_out
+  cerise.first_search.tsv
+  cerise.new_queries.fasta
+  cerise.second_search.tsv
+  command_logs.txt
+  $ grep '^>' cerise_out/cerise.new_queries.fasta | cut -f1 -d' ' | sort | diff - expected_new_queries__clustered_queries.txt
+  $ sort -k1,2 cerise_out/cerise.first_search.tsv | cut -f1,2 | diff - expected_first_search__clustered_queries.tsv
+  $ sort -k1,2 cerise_out/cerise.second_search.tsv | cut -f1,2 | diff - expected_second_search__clustered_queries.tsv
+
+Premade target DB; just targets clustered.
+
+  $ if [ -d cerise_out ]; then rm -r cerise_out; fi
+  $ mmseqs createdb clustered_targets.fasta clustered_targets.db > /dev/null 2>&1
+  $ cerise clustered_queries.fasta clustered_targets.db --target-clusters target_clusters.tsv --all-targets targets.fasta --search-config='-s=1 --threads=4' > cerise_oe 2>&1
+  $ ls cerise_out
+  cerise.first_search.tsv
+  cerise.new_targets.fasta
+  cerise.second_search.tsv
+  command_logs.txt
+  $ grep '^>' cerise_out/cerise.new_targets.fasta | cut -f1 -d' ' | sort | diff - expected_new_targets__clustered_targets.txt
+  $ sort -k1,2 cerise_out/cerise.first_search.tsv | cut -f1,2 | diff - expected_first_search__clustered_targets.tsv
+  $ sort -k1,2 cerise_out/cerise.second_search.tsv | cut -f1,2 | diff - expected_second_search__clustered_targets.tsv
+
+Queries have no hits in the DB.
+
+  $ if [ -d cerise_out ]; then rm -r cerise_out; fi
+  $ cerise silly_queries.faa clustered_targets.fasta --target-clusters target_clusters.tsv --all-targets targets.fasta --search-config='-s=1 --threads=4' > cerise_oe 2>&1
+  [1]
+  $ cat cerise_oe
+  ERROR: there were no hits of your queries to your targets
+  $ ls cerise_out
+  cerise.first_search.tsv
+  command_logs.txt

cerise/test/mmseqs_search.t/silly_queries.faa

@@ -0,0 +1,4 @@
+>s1
+xxa
+>s2
+VVVVVVVVVVVVVVV