RM signatures from tomtom

docs/tutorials/Tutorial_D2_TOMTOM.ipynb

@@ -26,7 +26,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 10,
+   "execution_count": 1,
    "id": "29dff7e7-ca13-4967-bf5c-2e9385a73571",
    "metadata": {},
    "outputs": [
@@ -36,7 +36,7 @@
        "(1, 1646)"
       ]
      },
-     "execution_count": 10,
+     "execution_count": 1,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -72,7 +72,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 13,
+   "execution_count": 2,
    "id": "2a66616d-457c-4d25-8eb0-f831309aa415",
    "metadata": {},
    "outputs": [
@@ -84,7 +84,7 @@
        "       'MA1332.1 BEH2', 'MA1333.1 BEH3', 'MA1493.1 HES6'], dtype='<U28')"
       ]
      },
-     "execution_count": 13,
+     "execution_count": 2,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -105,7 +105,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 16,
+   "execution_count": 3,
    "id": "eed69064-ebf5-4786-89fe-c574a224ddc9",
    "metadata": {},
    "outputs": [
@@ -115,7 +115,7 @@
        "(12, 1646)"
       ]
      },
-     "execution_count": 16,
+     "execution_count": 3,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -137,7 +137,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 17,
+   "execution_count": 4,
    "id": "e26732ac-31c3-411b-a785-611d45cae92e",
    "metadata": {},
    "outputs": [
@@ -149,7 +149,7 @@
        "       -2.18491891e-13,  2.02698924e-09, -1.84918747e-12,  1.90786831e-09])"
       ]
      },
-     "execution_count": 17,
+     "execution_count": 4,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -174,15 +174,15 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 19,
+   "execution_count": 5,
    "id": "80e8d169-bf4f-43ba-9a61-57776bd7ddbe",
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "1.22 s ± 71.3 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
+      "1.06 s ± 56.6 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
      ]
     }
    ],
@@ -202,15 +202,15 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
+   "execution_count": 6,
    "id": "4228e04b-03ff-4233-b5e0-1c5be2b61397",
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "10.1 s ± 222 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
+      "9.09 s ± 121 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
      ]
     }
    ],
@@ -238,7 +238,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 23,
+   "execution_count": 7,
    "id": "398c45b7-6540-48c7-9446-f75ece60d1a7",
    "metadata": {},
    "outputs": [
@@ -248,7 +248,7 @@
        "(1646, 100)"
       ]
      },
-     "execution_count": 23,
+     "execution_count": 7,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -268,7 +268,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
+   "execution_count": 8,
    "id": "05d713c5-5226-4f1a-924d-756ac0b4ef2f",
    "metadata": {},
    "outputs": [
@@ -279,7 +279,7 @@
        "        805.])"
       ]
      },
-     "execution_count": 24,
+     "execution_count": 8,
      "metadata": {},
      "output_type": "execute_result"
     }

tangermeme/tools/tomtom.py

@@ -11,7 +11,7 @@
 from numpy import uint64
 
 
-@njit("float64(float64[:], float64[:, :], float64, float64, int64[:])")
+@njit
 def _binned_median(x, bins, x_min, x_max, counts):
 	"""An internal function for calculating medians quickly.
 
@@ -49,9 +49,7 @@ def _binned_median(x, bins, x_min, x_max, counts):
 	return -99999
 
 
-@njit('''uint64(float64[:, :], float64[:, :], int32[:, :], float64[:, :], 
-	float64[:, :], float64[:], float64[:, :], float64[:], float64[:], int64[:], 
-	int64, int64, int64)''')
+@njit
 def _integer_distances_and_histogram(X, Y, gamma, f, Z, medians, median_bins, 
 	X_norm, Y_norm, Y_counts, nq_csum, nq, n_bins):
 	"""An internal function for integerized scores and the histogram.
@@ -113,7 +111,7 @@ def _integer_distances_and_histogram(X, Y, gamma, f, Z, medians, median_bins,
 	return uint64(-i_min * bin_scale)
 
 
-@njit('void(float64[:], float64[:], float64[:], float64[:], int64)')
+@njit
 def _pairwise_max(x, y, y_csum, z, n):
 	"""An internal function for the pdf of the maximum of two pdfs.
 
@@ -138,8 +136,7 @@ def _pairwise_max(x, y, y_csum, z, n):
 			z[i] = x[i] * y_csum[i] + y[i] * x_csum - x[i] * y[i]
 
 
-@njit('''void(float64[:, :], float64[:, :, :], float64[:, :], float64[:, :, :], 
-	int64, int64, int64, uint64)''')
+@njit
 def _p_value_backgrounds(f, A, B, A_csum, nq, n_bins, t_max, offset):
 	"""An internal function that calculates the backgrounds for p-values.
 
@@ -213,8 +210,7 @@ def _p_value_backgrounds(f, A, B, A_csum, nq, n_bins, t_max, offset):
 			B[i, j] = 1 - B[i, j]
 
 
-@njit('''void(int32[:, :], float64[:, :], uint64[:], int64[:], int64, uint64, 
-	int64, float64[:, :])''')
+@njit
 def _p_values(gamma, B_cdfs, rr_inv, T_lens, nq, offset, n_bins, results):
 	"""An internal function for calculating the best match and p-values.
 
@@ -261,7 +257,7 @@ def _p_values(gamma, B_cdfs, rr_inv, T_lens, nq, offset, n_bins, results):
 		total_offset += nt
 
 
-@njit('void(float64[:, :])')
+@njit
 def _merge_rc_results(results):
 	"""An internal method for taking the best across two strands."""
 
@@ -282,9 +278,7 @@ def _merge_rc_results(results):
 			results[i, 4] = 1
 
 
-@njit('''float64[:, :, :](float64[:, :], float64[:, :], int64[:], int64[:],
-	float64[:], float64[:], uint64[:], int64[:], int64, int64, int64, int64, 
-	int64)''', parallel=True)
+@njit(parallel=True)
 def _tomtom(Q, T, Q_lens, T_lens, Q_norm, T_norm, rr_inv, rr_counts, n_nearest, 
 	n_score_bins, n_median_bins, n_cache, reverse_complement):
 	"""An internal function implementing the TOMTOM algorithm.