The local-resemblance map of two polyproteins

Alignment	Identities, %	Positives, %	Length	Gaps	Score	Score, bits
Range 1: 1684 to 2327	29	48	615, 644	59	638	250
Range 2: 1121 to 1383	37	51	277, 263	26	397	157
Range 3: 298 to 718	26	40	486, 421	81	307	122
Range 4: 811 to 917	23	40	121, 107	28	82	36,2
Range 5: 554 to 637	24	41	83, 84	14	46	22,3
Range 6: 1810 to 1837	36	53	24, 28	4	45	21,9
Range 7: 442 to 459	39	60	23, 18	5	41	20,4
Range 8: 1237 to 1262	31	48	29, 26	3	41	20,4
Range 9: 224 to 235	58	75	12, 12	0	40	20,0
Range 10: 894 to 929	26	55	38, 36	2	38	19,2

The best Score results are 250 and 157 bits.
For the first result: PO330 - Protein 3CD P03306 - RNA-directed RNA polymerase 3D-POL (1863 - 2332), Picornain 3C (1650 (1684) - 1862;
probably, there are one protein and the piece of other protein in this alignment
For the second result: P0330 - Protein 2C (probably, the part of rhe protein) Po3306 - Protein 2C (probably, the part of rhe protein)

Comparing the weight of alignment with random

ID1, ID2	Score	Median (m)	Upper quartile (Q1)	Bits (B)	Probability (P)
TRPB_ECOLI, TRPB_BACSU	1102.0	55.25	62.25	149.7	8.626E-46
DNAA_ECOLI, FADA_ECOLI	55.5	45.0	50.0	2.3	0.203

So, we can see that the first alignment is not accidentally; and the second may be appeared because of the occasional reasons

Increasing of penalty for gap extension (= 4.0)

ID1, ID2	Score	Median (m)	Upper quartile (Q1)	Bits (B)	Probability (P)
TRPB_ECOLI, TRPB_BACSU	1100	36	38.00	532.5	4.03E-161
DNAA_ECOLI, FADA_ECOLI	44	34	37	3.7	0.077

So, we can notice that increasing of penalty for gap extension (= 4.0)
gives us decrease of the probability of accidental alignment

Alignments using BLOSUM30

ID1, ID2	Score	Median (m)	Upper quartile (Q1)	Bits (B)	Probability (P)
TRPB_ECOLI, TRPB_BACSU	1498	297.75	308.75	109.2	1.34E-33
DNAA_ECOLI, FADA_ECOLI	264	277.5	289.25	-1.06 (uncorrect)	2.08 (uncorrect)

With the help of BLOSUM30 we got results with higher significans of probability (P).
We can also see negative bit values, which are incorrect (even with the recount of score).
Thus, we can say that BLOSUM30 does not always work properly

Checking the formula for conversion to bits

Let's check our results using standard parameters.
In the first case we will look at trpb_ecoli and trpb_bacsu and in the second case - dnaa_ecoli and fada_ecoli.
We will shuffle trp_bacsu and fada_ecoli sequences for 1000 terms

ID1, ID2	Upper Quartile (Q1)	Median (m)	Upper 1/8	Bits (B)	Probability (P)
TRPB_ECOLI, TRPB_BACSU	61.25	54.5	67.5	2.1	0.233
DNAA_ECOLI, FADA_ECOLI	50.5	45.0	55.0	2	0.25

The bits'results smaller than 3 for a while

BLAST: search for homologs in the database